Machining Shaft on CNC: How Geometry, Setup and Measurement Decide Accuracy

• 1
  What machining shaft work really asks of the machine
• 2
  Diameter, roundness and runout are three different numbers
• 3
  Where slenderness changes the process plan
• 4
  How setup choice moves the centerline
• 5
  Material choice changes the cutting window
• 6
  Measuring a shaft so the numbers mean something
• 7
  Choosing the operation for each shaft feature

A shaft is a long part that rotates. That single fact drives everything else. The diameter can be measured with a micrometer, but the feature that matters is the axis: every journal, shoulder, keyway and thread must sit on one common centerline. If the centerline drifts, the part still measures in tolerance on diameter and still vibrates in the assembly.

When we plan machining shaft work, we start from the ratio of length to diameter. A part at 3:1 behaves. A part at 8:1 starts to deflect under cutting force and to whip at higher spindle speed. The same 0.5 mm depth of cut that is safe on a stubby part will push a slender one off center by several times the tolerance band before the tool exits.

The second driver is how the part will be held. Between centers, in a chuck with a tailstock, or on a fixture plate: each choice adds its own error stack. Chuck jaws add runout. A tailstock adds thrust. A steady rest adds a contact point that can mark the surface. None of these are reasons to avoid turning a shaft. They are reasons to state the datum clearly on the drawing.

Finally, the shaft is rarely finished on one machine. Journals get ground, keyways get milled, threads get rolled or cut, and surfaces get plated or anodized. Every operation after turning can move the centerline. The engineering question is not whether the part can be turned. It is which operations you allow after turning, and how much of the tolerance you spend on each.

Engineers often send one callout and expect three properties. Diameter is a size. Roundness is a form. Runout is a relationship between the surface and an axis. A turned journal can be perfectly round and still have high runout if the centers were drilled off axis. It can also be on size and out of round if the part flexed during the finish pass.

Our standard machining tolerance is ±0.005 mm on critical diameters, and we hold Ra 0.8–1.6 μm as a normal turned finish. Finer than Ra 0.8 μm on a shaft usually means a grinding or burnishing step, not a slower turning pass. Pushing a single-point tool to a mirror finish on a long shaft tends to introduce chatter, which trades finish for roundness.

Roundness is the number that decides bearing life. A journal that is 0.01 mm out of round will load the rolling elements unevenly on every revolution. That is why bearing-seat drawings usually specify circularity separately from size, and why we measure roundness on a dedicated instrument rather than inferring it from a micrometer reading.

Runout is the number that decides vibration. It is set by the setup, not by the tool. If you drill centers in the same setup that turns the journals, runout stays small. If you re-chuck the part between operations, expect runout to grow with each re-clamp. A drawing that calls 0.005 mm total runout on a re-chucked shaft is asking for an operation the process cannot support reliably.

• 1
  SizeMicrometer or gauge, measured across the diameter at several stations.
• 2
  FormRoundness and cylindricity, measured against the part axis, not a V-block.
• 3
  RelationshipRunout and concentricity, referenced to a stated datum, not to the chuck.

The length-to-diameter ratio is the fastest way to predict whether a shaft will be straightforward or difficult. Below 4:1, a three-axis lathe with a tailstock handles most work in one or two setups. Between 4:1 and 8:1, deflection becomes the dominant error source and you need center support, lower cutting forces and a lighter finishing pass.

Above 8:1, the part stops behaving like a stiff beam and starts behaving like a spring. Cutting force bends it away from the tool, so the tool removes less material than programmed. The result is a barrel-shaped diameter: small at the ends, large in the middle, or the reverse depending on where the support sits. No amount of tool wear compensation fixes a stiffness problem.

For long shafts we prefer to reduce the number of interrupted cuts, keep the tool nose radius consistent, and take the finishing pass with a small depth of cut at a higher surface speed. If the geometry allows, we will also break the part into shorter machined sections joined by a designed feature rather than fight a 12:1 unsupported span.

There is a practical ceiling here. We machine up to 4,000 mm maximum processing size, and long slender parts are the case where that envelope is genuinely useful. But length alone is not the constraint. A 900 mm shaft at 20 mm diameter is harder than a 900 mm shaft at 80 mm diameter, and the quote should reflect that difference.

Between-centers turning is the most accurate way to keep a shaft on one axis, because the part is located by its own centers rather than by a chuck. The trade-off is that the centers themselves must be drilled concentric to the journals, and they must survive the whole process. If a later operation removes a center, the reference is gone.

Chuck-and-tailstock setups are faster and allow more of the part to be machined in one pass. The cost is runout. A three-jaw chuck will typically show 0.02–0.05 mm of runout at the far end of a short shaft, and more as length grows. Soft jaws bored in place reduce this, but they must be re-bored for each diameter change.

On our mill-turn centers we can turn, mill and drill a shaft without re-clamping, which removes an entire re-chuck error from the stack. That matters most for parts with a keyway and a journal on the same datum. It does not remove the need for a stated datum on the drawing. A part with no datum reference is measured against whatever the inspector decides, and that is how disputes start.

For production runs we will often make a soft-jaw set or a dedicated fixture after the first article passes. The fixture costs money once and then holds the datum for every part after it. On a 10,000-piece run that is the difference between a process that drifts and one that does not.

The same shaft geometry behaves differently in 1045 steel, 17-4PH stainless and 7075 aluminium. Carbon steel turns cleanly and holds size, but it will rust in storage if it is not protected. Stainless work-hardens at the surface if the tool rubs instead of cutting, so we keep the feed per tooth above the work-hardening threshold and never dwell.

Aluminium shafts are common in automation and robotics because the weight matters more than the strength. 6061-T6 and 7075 machine fast and hold a good finish. They also scratch easily, so an anodized finish is often specified not for corrosion but for wear at the bearing seat. Hardcoat anodizing adds a dimension change that must be accounted for before grinding.

Titanium such as TC4 (Ti-6Al-4V) is used where weight and corrosion resistance both matter, usually in aerospace or medical work. It cuts at low surface speed with high coolant pressure, and it springs back against the tool. Thin titanium shafts are the hardest case in our shop, and we would rather grind than turn the final journal.

Plastics and PEEK appear in medical and electronics shafts where the part must be non-magnetic or chemically inert. Here the enemy is heat, not force. A sharp tool, a high rake angle and air or mist cooling keep the diameter stable. PEEK holds size well once it is stress-relieved; ABS and PP do not, and a tight tolerance on a PP shaft is usually a design error.

• 1
  Carbon and alloy steelPredictable turning; protect from corrosion after machining.
• 2
  Stainless 303, 304, 316, 17-4PHControl feed to avoid work hardening at the surface.
• 3
  Aluminium 6061, 7075Fast, light, needs anodizing for wear at the seat.
• 4
  Titanium and plasticsLow cutting speed, high coolant, expect springback or heat growth.

A shaft is measured against its own axis. That sounds obvious and is often ignored. A micrometer reading taken at three points along a journal tells you about size and taper. It says nothing about roundness or runout. Inspection of a shaft therefore needs at least three instruments or three setups: a size gauge, a roundness reference and a runout reference.

We inspect 100% of parts before shipment, with raw material check, in-process monitoring and final inspection, and we can supply reports on request. On a shaft the in-process check matters most. If the first finishing pass shows 0.004 mm of taper, the operator can adjust before the whole batch is cut. If the same taper is found at final inspection, the batch is scrap.

For concentricity, the datum must be a feature that survives the process. A center hole is a good datum because it was used to locate the part. A chucked diameter is a poor datum because it was created by the same setup that produced the error. When a drawing calls concentricity to a chucked surface, we ask the engineer to move the datum to the center hole or to a ground journal.

Temperature is the quiet variable. A steel shaft 300 mm long grows about 0.003 mm for every 1 °C of temperature change. If the part is measured hot off the grinder and compared with a part measured at 20 °C in the inspection room, the numbers will not match. We let parts settle before final measurement on tight-tolerance work.