Professional Machinery and CNC Precision: How Accuracy Is Actually Made
This page explains what separates professional machinery and CNC precision work from general job-shop machining. It is written for design and manufacturing engineers who need to judge whether a quoted tolerance is repeatable, and where the real accuracy limits sit.
What the Fifth Axis Changes in the Cut
A three-axis mill moves the tool in X, Y and Z. The part stays still, so every feature that faces away from the spindle needs a second setup. Each setup adds a work offset, a re-clamp and a fresh chance for error. Professional machinery and CNC precision work starts by counting those setups, because they are usually the largest single contributor to a stack-up.
Adding two rotary axes lets the tool reach the part from an angle instead of the part being turned to face the tool. On a simultaneous five-axis center, the A and B (or B and C) axes move while the tool is in the cut. That keeps the cutting point and the tool axis under continuous control, which is what allows a single continuous pass over a curved surface.
The practical gain is not only reach. A short, stiff tool can be kept normal to the surface for the whole pass. Tool overhang drops, deflection drops, and the surface comes off more even. On a deep cavity, a three-axis machine may need a long slender tool that chatters; a five-axis machine tilts the head and uses a stubby tool instead.
The trade-off is that rotary axes bring their own error. Rotary positioning, pivot distance and centerline offset must be calibrated, and they drift with temperature. So a five-axis machine does not automatically hold a tighter tolerance than a good three-axis machine. It removes setups. That is a different benefit, and it should be quoted as such.
- 1Fewer setupsOne fixturing instead of three or four, so datum error stops compounding.
- 2Shorter toolsTilted access allows stiffer tooling and less chatter on deep features.
- 3Continuous passesSurfaces cut in one path avoid witness lines between blended regions.
Where Tolerance Really Comes From
A drawing tolerance of ±0.005 mm is a statement about the finished part, not about the machine. The machine contributes only one term. Fixture location, material condition, coolant strategy, tool wear and thermal state all add to the total. When a shop quotes a tight number, ask which of these terms it controls and how it verifies them.
Thermal drift is the term most often ignored at the quoting stage. A spindle running at 12,000 rpm heats up over the first two hours. A 2 °C rise across a 400 mm aluminum part moves it roughly 0.009 mm before any cutting error is counted. That is larger than the tolerance itself. Shops that hold tight numbers run warm-up cycles, keep the room at controlled temperature, and measure after the part has stabilized.
Material behavior matters just as much. Aluminum 6061 and 7075 cut cleanly and hold a fine finish. Titanium Ti-6Al-4V and Inconel work-harden at the cut edge, generate far more heat, and spring back after the tool passes. A 0.05 mm finishing allowance on Inconel may not clean up in one pass, and thin walls will deflect under the cutting force. These are process decisions, not machine specifications.
Inspection closes the loop. A coordinate measuring machine at 20 °C gives a number that is traceable only if the part is at the same temperature. Laser scanning adds coverage for free-form surfaces but needs its own alignment strategy. For a part with a true position callout, the fixture that held it during machining should also hold it during inspection, or the datum reference frame changes between the two operations.
- 1Cutting forceThin walls deflect during the pass and spring back after it.
- 2Tool wearA worn edge changes effective diameter and pushes the size out over a run.
- 3ClampingOver-tight vises distort a bore that measures round only after release.
When a Five-Axis Machine Is the Wrong Answer
Simultaneous five-axis motion is expensive in programming and cycle time. A part that is essentially a plate with holes, pockets and a flat back face should stay on a three-axis machine. Two setups on a three-axis mill can cost less than one five-axis cycle, and the tolerance outcome is the same. Reach and setup count decide this, not the axis count on the spec sheet.
Prismatic parts with a single critical face are another case. If the datum face and the critical face can be machined in one three-axis setup, adding rotary axes only adds calibratable error sources. The same logic applies to small prototype quantities where programming time dominates the total. For one or two pieces, a well-fixtured three-axis part often ships sooner.
Five-axis earns its place when the geometry is genuinely free-form, when a feature sits on many faces at different angles, or when the tolerance chain crosses several datums. Impellers, turbine blades, bone plates with compound curvature, and injection mold cores with deep ribs are the classic fits. So are housings where six sides need machined pads held to one another.
Size is the other boundary. A part that fits a 500 × 500 × 450 mm envelope is routine. As the work envelope grows toward 4,000 mm, thermal and geometric error grow with it, and the choice of machine matters more than the axis count. Large structural parts often run better on a large three-axis or mill-turn platform with careful fixturing than on a five-axis machine near its travel limit.
- 1One dominant faceThree-axis with a solid fixture is faster and just as accurate.
- 2Free-form surfacesContinuous five-axis passes remove blend lines and hand polishing.
- 3Many angled facesOne five-axis setup replaces a stack of fixtures and re-datums.
How a Shop Holds ±0.005 mm Repeatably
Repeatability is a system property, not a machine property. It comes from controlling the inputs on every run: the same tool holders, the same warm-up routine, the same probe cycle, the same inspection method. When one of those changes between runs, the size distribution widens even if the machine is unchanged.
In-process probing is the most direct control. After roughing, the probe measures the actual stock left and the control adjusts the finishing pass. This absorbs material variation in castings and forgings, which can move the stock by a millimeter or more. It also catches a fixture that was loaded with chips under a locating pad.
Tool life management keeps the finishing pass predictable. A carbide end mill that has cut 40 minutes of aluminum is not the same tool that started. Scheduling a tool change before the wear land reaches the point where size drifts is cheaper than scrapping a near-finished part. For long runs, a spare pre-set tool in the magazine removes the setup delay.
Finally, the report matters. A dimensional report produced from a defined inspection plan, with the datum scheme stated, is what lets a purchasing engineer accept a lot without re-measuring every part. GreatLight runs raw material checks, in-process monitoring and a 100% inspection before shipment, with reports available on request.
- 1Warm-up firstRun the spindle and axes to thermal steady state before the first finish pass.
- 2Probe the stockMeasure before finishing so the control can adapt to real material condition.
- 3Change tools earlyScheduled tool changes cost less than a scrapped near-finished part.
Material Choice Sets the Achievable Finish
Surface finish and tolerance are linked through the material. Aluminum and brass cut freely and reach Ra 0.2–0.8 μm with a clean finishing pass. Stainless 316 work-hardens and tends to smear, so a slightly coarser Ra 0.8–1.6 μm is a realistic target without a secondary operation. Titanium and Inconel sit at the demanding end of the range.
The as-machined band of Ra 1.6–3.2 μm covers most functional surfaces: bearing bores that will be honed later, mounting faces, and general structural interfaces. Specifying a finer finish than the function needs adds cycle time and often forces a second operation that introduces a new datum. It is worth asking what the surface actually does before tightening the callout.
Finishing operations change dimensions unless they are planned for. Anodizing builds a coating on the surface, so a hardcoat layer will shift a press-fit bore. Electroless nickel and plating do the same. Bead blasting rounds edges slightly and can close a small chamfer. If a critical dimension sits on a coated surface, the pre-coat size has to be adjusted at the CAM stage, not after.
Some features cannot survive a coating at all. Threads, dowel holes and sealing faces are usually masked. Laser marking needs a minimum character height of about 1.5 mm to stay legible after anodizing, so small parts may need the marking moved to a flat, uncoated area.
- 1Aluminum and brassReach Ra 0.2–0.8 μm with a single finishing pass.
- 2Stainless and titaniumExpect Ra 0.8–1.6 μm without a secondary polishing step.
- 3Plated partsAdjust pre-plate size; mask threads and dowel holes.
Three-Axis vs Four-Axis vs Five-Axis: Which Fits the Part
Match the machine to the geometry and the setup count, not to the brochure.
| Part characteristic | 3-axis | 4-axis | 5-axis |
|---|---|---|---|
| Flat plate, holes, pockets | Best fit | Not needed | Overkill |
| Deep cavity, single face | Good with long tool | No gain | Better with tilted stub tool |
| Features on 4+ faces | 3-4 setups | 2 setups | 1 setup |
| Free-form curved surface | Blend lines likely | Partial reach | Continuous pass |
| Impeller, blade, bone plate | Not practical | Limited | Best fit |
| Tight true position across datums | Setup error adds up | Fewer re-datums | Lowest setup error |
| One or two prototype pieces | Fastest to program | Rarely worth it | Programming dominates |
| Part near 4,000 mm long | Large platform | Rare | Travel limit risk |
The Practical Verdict
If the part needs three or more setups, or has free-form surfaces and angled features, use a five-axis machine. If it is a prismatic part with one dominant face, a well-fixtured three-axis setup is cheaper and just as accurate.
Questions Engineers Ask
Does a five-axis machine hold a tighter tolerance than a three-axis machine?
Not by itself. The machine geometry and the rotary calibration set the floor, and that floor is often similar to a good three-axis machine. The advantage is setup reduction. With one fixturing, datum error stops compounding across operations, so the finished part is more likely to land inside a tight true position callout.
If a shop claims a tighter tolerance purely because the machine has five axes, ask for the inspection plan. The number that matters is the one measured on the finished part, at a stated temperature, against a stated datum scheme.
How does thermal drift affect a ±0.005 mm callout?
Steel and aluminum expand roughly 11–23 μm per meter per °C. A 2 °C shift across a 400 mm part moves it about 0.009 mm, which is already larger than the tolerance. That is why warm-up cycles, controlled shop temperature and measurement after stabilization are part of holding the number, not optional extras.
For long parts, the effect scales with length. A 1,000 mm aluminum part can move more than 0.02 mm from a modest temperature change, so the inspection temperature has to be stated on the report.
When is a second setup cheaper than a five-axis cycle?
When the second face is flat and easy to locate. Two setups on a three-axis machine with a simple vise and a stop add a small amount of re-datum error but almost no programming cost, and the cycle time is shorter. For a plate with a machined front and back, that is usually the cheaper route.
The comparison flips once the part needs four or more faces, or when the second face carries a feature that must align to the first within a tight true position. At that point the fixture stack costs more than the five-axis cycle.
Which materials are hardest to hold to a tight tolerance?
Titanium Ti-6Al-4V, Inconel and thin-wall aluminum are the usual answers. Titanium and Inconel work-harden at the cut edge and hold heat, so the tool wears quickly and the part springs back after the pass. Thin-wall aluminum deflects under cutting force and measures round only after the vise is released.
Magnesium AZ31B and AZ91D cut freely but need chip control and careful handling. Plastics such as PEEK and POM move with temperature and moisture, so they are often machined with a finishing allowance and a stress-relief step before the final pass.
What does 100% inspection actually cover?
It means every part is checked against the drawing before shipment, not that every dimension on the drawing is measured on every part. The plan defines which characteristics are measured on all parts and which are sampled, plus the gauge and the datum scheme used.
At GreatLight the sequence is raw material check, in-process monitoring, then final inspection. Reports are available on request, and for coated parts the critical dimensions are checked after the coating, since plating and anodizing shift the size.
Can a prototype and a 10,000-part run use the same process?
They can share the same datums and the same inspection plan, which is what keeps the transition clean. The cutting strategy usually differs. A prototype may be machined from billet with generous stock and a single finishing pass. A production run uses fixtures, probing and scheduled tool changes to hold the same size across thousands of parts.
There is no minimum order quantity at GreatLight, so a program can start with one piece and scale to 10,000+ without changing the datum scheme on the drawing.
Send the Drawing, Get a Process Answer
Upload your model and we will return a quotation with a free DFM analysis within 12 hours, including a note on where the tolerance is likely to be hardest to hold.
12-hour quote100% inspectionNo minimum order quantityNDA on request
More Machining Notes
We publish setup notes, tooling trials and inspection data from the factory floor.
