CNC Precision Processing Guide: How Tolerance, Setup and Material Decide Your Part

• 1
  What precision actually means on a machine tool
• 2
  Why setup count drives both accuracy and cost
• 3
  How material choice changes the achievable tolerance
• 4
  Surface finish, tolerance and inspection are linked
• 5
  Which machine type fits the geometry

Precision is not one number. It is the sum of machine positioning error, thermal drift, tool wear, fixture stiffness and the metrology used to check the result. A machining center rated at ±0.005 mm can hold that on a 40 mm bore in aluminium, but the same machine may only reach ±0.02 mm on a 900 mm long stainless shaft after four hours of cutting. The rating is a capability under defined conditions, not a promise for every geometry.

Thermal growth is the quiet one. A spindle running at 12,000 rpm warms the headstock by several degrees over the first hour, and a 500 mm steel part grows roughly 0.006 mm per °C. Shops that hold tight tolerances either warm the machine up before the first cut, run coolant at a controlled temperature, or schedule the critical finishing pass after the machine has stabilized.

Tool wear moves the cut in one direction. On a 6 mm carbide end mill in 304 stainless, flank wear of 0.02 mm is normal after 30–40 minutes in the cut, and that error goes straight into the part unless the operator offsets the tool or swaps it before the finishing pass. This is why a tight tolerance on a long run is a tool management problem, not only a machine problem.

Every time a part is unclamped and re-fixtured, a new datum error enters the stack. A three-axis job that needs five faces machined may need three or four setups, and each one adds a small but real positional shift. Five-axis machining removes most of that. With 16 simultaneous 5-axis centers in our shop, a part can be finished from one orientation and the datum error stays in the machine, not in the fixture.

The benefit is not only accuracy. Fewer setups mean less queue time between operations and fewer chances for a burr or a clamp mark on a finished face. For parts with angled holes, contoured pockets or features on multiple sides, five-axis is usually the cheaper route once you count the total hours, even though the hourly rate is higher.

There is a limit. Five-axis is not worth it for a flat plate with holes on one face. A three-axis machine with a good vise will run that faster and cheaper. The decision point is geometric: count the number of distinct tool approach directions the drawing needs. One direction is three-axis work. Two or three directions on a part with tight positional callouts is where rotary or five-axis starts to win.

Fixturing still matters on five-axis. Thin walls deflect under clamping force, and a 1.5 mm aluminium wall will move more than the tolerance allows if it is gripped too hard. Soft jaws, vacuum plates or sacrificial tabs are common answers. The rule we use: if the part moves when you push it with a thumb, the cut will move more.

Aluminium 6061-T6 cuts clean and holds ±0.005 mm on most features. It also moves after machining if the stock was not stress-relieved, especially on thin sections. For a 200 mm long bracket with a 3 mm web, we often rough, let it rest, then finish. That extra step costs time but keeps the part flat.

Stainless 304 and 316 work-harden. A dull tool rubs instead of cutting, the surface hardens, and the next pass cuts worse. This is why 304 parts often show a better finish from a fresh insert at moderate speed than from a worn tool pushed hard. 17-4PH behaves differently again and is usually machined in the solution-treated condition, then aged.

Titanium Ti-6Al-4V and Inconel are the slow ones. They conduct heat poorly, so the cutting edge runs hot and tool life drops fast. Tolerances are still achievable, but cycle time may be three to five times that of aluminium for the same feature. If a design can use 7075 instead of Ti-6Al-4V and still meet the load case, the saving is large.

Plastics are a separate case. POM and PEEK machine well but move with temperature and moisture. A PEEK part measured right off the machine can be 0.05 mm different the next morning. For tight plastic work, we machine, let it condition, then take the final cut. ABS and PC are softer and tend to burr, so deburring becomes part of the tolerance budget.

A tight tolerance and a fine finish are not the same requirement, but they interact. Ra 0.2–0.8 μm usually needs a finishing pass with a small stepover and a sharp tool. That pass removes very little material, so if the roughing left 0.05 mm of stock, the finish pass may not clean up. The result is a part that measures in tolerance but shows tool marks.

Ra 0.8–1.6 μm is the common engineering finish and covers most sealing faces and bearing bores. Ra 1.6–3.2 μm is fine for non-critical surfaces and costs less. Specifying Ra 0.2 μm on a face that only needs to look clean adds cycle time for no functional gain.

Inspection closes the loop. We check raw material certificates on arrival, monitor in-process dimensions, and inspect 100% of parts before shipment, with reports on request. For a first article, a CMM report showing actual values against the drawing is the cheapest way to catch a datum error before a full run.

One practical point on drawing callouts. A general tolerance block of ±0.1 mm with three tight features called out individually is easier to make than a blanket ±0.02 mm on every dimension. The machinist will chase the tight ones and leave the rest alone. Blanket tight tolerances on a large part mostly add cost.