CNC Start Successfully: What Happens Before the First Chip

• 1
  Key takeaways
• 2
  What It Means to CNC Start Successfully
• 3
  How the Setup Chain Determines the First Cut
• 4
  Workholding Choices and Their Real Limits
• 5
  How Material Condition Changes the Parameters
• 6
  First-Article Inspection and What It Proves
• 7
  When a CNC Start Is the Wrong Tool
• 8
  Which Setup Fits Which Part

A CNC start is the window between loading a program and cutting a part that passes inspection. Everything before that window is preparation: reading the drawing, choosing stock, planning datums, building the fixture, and proving the tool path. When engineers ask how to CNC start successfully, they usually mean how to remove uncertainty from that window.

The reason it matters is cost. A scrapped first article on a five-axis job ties up a spindle, a fixture, and an inspection slot. A scrapped first article on a 10,000-part run is worse, because the error is baked into a process that already looked approved. A clean CNC start is cheaper than a fast one.

There is a common assumption that modern CAM software removes the risk. It does not. CAM calculates a path from a nominal model. The machine cuts real stock that may be 0.3 mm oversize, clamped in a fixture that deflects under load, with a tool that wears from the first cut. A CNC start successfully executed means those real conditions were measured and compensated before the finish pass.

Every machining operation inherits position from a previous one. If the first operation establishes a face and two edges as datums, the second operation locates from those features. Error compounds along the chain. On a part with a ±0.005 mm position callout, the budget is consumed quickly, so the setup plan has to place the tightest tolerances in the fewest possible setups.

This is why five-axis work changes the plan. On a simultaneous 5-axis center, the part can be cut from five sides in one setup, which removes re-fixturing error entirely. That matters for parts with compound angles, deep pockets, or features that must stay coaxial. On simpler prismatic parts, three-axis work with two setups is often faster and cheaper, as long as the datum transfer is controlled.

Thermal drift is the quiet variable. A spindle warming from cold to steady state moves a few micrometres over the first hour. Shops that start a tight-tolerance job cold and measure at minute five are measuring a different machine than the one that runs at minute sixty. Warm-up cycles and in-process probing are the usual countermeasures.

• 1
  Datum hierarchyChoose primary, secondary, and tertiary datums that survive every operation, and avoid using a feature that will later be removed.
• 2
  Setup countEach re-fixture adds position error. Consolidate features into one setup when the tolerance is tighter than ±0.02 mm.
• 3
  Stock allowanceLeave enough material for the finish pass, typically 0.3–0.5 mm on faces and 0.15–0.25 mm on walls.

A vise is the default for prismatic parts because it is fast and repeatable. Its limit is jaw lift and part distortion on thin walls. When a part is clamped with 2,000 N of force, a 2 mm wall will deflect; the cut removes material from a deflected shape, and the part springs back out of tolerance after unclamping. Soft jaws machined to the part profile are the usual fix.

For thin plates and housings, vacuum fixturing holds the part flat with distributed force. The trade-off is low holding force, so cutting loads must stay light and the tool path must avoid lifting the part. For asymmetric parts, a machined fixture plate with locating pins and clamps gives repeatable position across a run and is worth the setup cost when more than a handful of parts are needed.

On mill-turn centers, bar stock can be fed and gripped by the spindle, so the part never leaves the machine between turning and milling. This eliminates a whole class of concentricity errors. It is the right choice for parts like shafts, fittings, and bushings where a turned diameter and a milled flat must share an axis.

Aluminium 6061-T6 cuts cleanly at high surface speed, but it is soft enough to build a built-up edge on the tool if the feed is too light. Too fine a chip load rubs instead of cuts, which raises heat and degrades finish. The practical range for roughing is a moderate feed per tooth that keeps the cutter engaged, with coolant or air blast to clear chips from pockets.

Stainless 304 and 316 work harden. If the tool dwells, the surface hardens under the cut and the next pass meets a harder skin. The countermeasure is constant engagement: keep the cutter moving, avoid stopping in the cut, and never take a spring pass on a hardened surface. Tool wear accelerates quickly once work hardening starts, so counting tools per part matters more than watching the clock.

Titanium Ti-6Al-4V and Inconel behave differently again. They hold heat in the cut rather than carrying it away in the chip, so the cutting edge runs hot even at low surface speed. High-pressure coolant directed at the edge, conservative depths of cut, and sharp tools with a positive rake are the standard approach. These materials also spring back after cutting, so a finish pass may need a small allowance and a second measurement.

A first article is not a sample. It is the full inspection of the first part off a new process, checked against every drawing callout that the process is expected to control. The measurement method has to match the callout: a caliper is fine for a general dimension but cannot verify a true position of ±0.005 mm or a surface finish of Ra 0.8 μm.

CMM inspection covers position, profile, and form. Surface finish requires a profilometer or a comparator. Threads need go/no-go gauges. When a shop issues a first-article report, the report should list the nominal, the tolerance, the measured value, and the instrument used. A report that only says 'pass' is not evidence.

For production runs, the first article is followed by in-process monitoring and a final inspection before shipment. The purpose is to catch drift, not to re-prove the setup. A process that was correct at part one and wrong at part five hundred has a wear or thermal cause, and that is what in-process checks are for.

CNC machining is subtractive, so it cannot make a part cheaper by adding material efficiency. For a hollow shell with thin, uniform walls, die casting or vacuum casting will beat machining on unit cost once volume justifies the tooling. CNC remains the right call for the first article and for the bridge quantity before tooling is ready.

Very soft or gummy plastics such as PP and HDPE can be machined, but they deflect under clamping and smear at the cut edge. They are workable, not ideal. For a part that must be transparent, PMMA and PC machine well but need polishing after, and the polished surface is not always uniform across a batch.

Geometry with internal channels that cannot be reached by a rotating tool is outside the process. So is a part whose smallest internal radius is smaller than the available cutter. In both cases, a design change or a different process such as 3D printing is the honest answer, and it is better to say so before a quote than after a failed first article.