CNC Production Plan: How a Part Moves From Quote to Shipped

• 1
  What a CNC production plan actually controls
• 2
  Datum schemes decide whether the tolerance is reachable
• 3
  Choosing the machine class from part geometry
• 4
  Where cycle time is actually lost
• 5
  Inspection gates belong inside the plan, not after it
• 6
  When a CNC production plan should be rejected
• 7
  Matching the plan to the part

A CNC production plan is not a G-code file. It is the decision layer above the code: which machine runs which feature, in what order, on what fixture, measured how, and released when. The code only executes what the plan already decided. When the sequencing is wrong, the machine will still cut metal, just not the right metal in the right place.

Think of it as a routing sheet with consequences. Every entry names an operation, a workholding method, a tool list, a target dimension and an inspection gate. Those entries set the tolerance budget before a single chip is made. Change one and you change the budget for everything downstream.

We build this layer during quotation and free DFM analysis, which we return within 12 hours. By the time the plan reaches the shop floor, the fixture concept, the datum scheme and the inspection method are already fixed. Production can start within 24 hours after approval.

Tolerance is a stack, not a single number. A ±0.005 mm callout on a bore means nothing until you know which face locates the part and which face the probe touches. If the datum printed on the drawing is not the surface the fixture actually clamps, the machine will chase a moving target all day.

For a bracket with a flat back and two dowel holes, the natural route is: face the back, drill the dowels, then use those dowels as the secondary datum for the front-side work. Two setups, one shared reference, no re-zeroing guesswork. That is the difference between a plan and a stack of operations.

Thin walls complicate this. Below roughly 2 mm wall thickness on aluminium, clamping pressure itself moves the part. The fix is usually a soft jaw machined to the finished profile, or a vacuum plate that spreads the load. Neither is exotic, but both must appear in the plan before the stock is cut.

Machine choice follows geometry, not habit. A part with features on five faces and a positional tolerance under ±0.02 mm belongs on a simultaneous 5-axis center, where the part stays in one setup. We run 16 of them. A shaft with a cross-hole and a keyway usually belongs on a mill-turn center, of which we run 16.

When the part fits inside 500 × 500 × 450 mm and needs three faces, a 3-axis machine with two fixtures is often cheaper than a 5-axis single setup. We run 27 three-axis machines and 12 four-axis mills for exactly this reason. The plan should name the class and justify it.

Size changes the calculus again. Our largest travel is 4,000 × 400 × 150 mm. Beyond that envelope, the plan has to split the part or move to fabrication. Pretending a large frame can be machined in one hit is how projects miss dates.

Cycle time is not dominated by the cut. On most parts under 300 mm, tool changes and air moves eat more clock than the finishing pass. A plan that groups features by tool, rather than by drawing order, can cut 15 to 30 percent off the cycle without touching feeds.

Deep pockets are the other trap. A pocket deeper than three times the tool diameter needs a smaller tool, a longer reach and a reduced feed. Those reductions compound. If the plan assumes the same feed as a shallow pocket, the quoted cycle time is fiction.

Hard materials push this further. Inconel and Ti-6Al-4V cut at a fraction of aluminium speeds and heat the tool faster. For those, the plan should budget more passes and shorter tool life, not just a slower spindle. A plan that ignores this ships late.

Inspection is a step, not a final formality. We place gates after first-article machining, after any heat treatment, and after finishing, because each of those can move a dimension. A coating that adds 20 μm changes a press fit. If nothing measures the part between those steps, the error is only discovered at the end.

Our standard flow is raw material check, in-process monitoring, then 100 percent inspection before shipment, with reports on request. The plan states which dimensions are checked at which gate and with what instrument. Calipers are fine for a ±0.1 mm slot. They are not fine for a ±0.005 mm bore.

Surface finish has its own gates. Ra 1.6–3.2 μm is a normal as-machined result. Ra 0.8–1.6 μm takes a controlled finishing pass. Ra 0.2–0.8 μm needs a deliberate sequence and often a different tool. The plan should say which band applies and how it will be verified.

Reject a plan that lists operations without naming fixtures. Reject one that quotes a tolerance tighter than the datum scheme can support. Reject one that has no inspection step between machining and finishing on a coated part. These are not stylistic complaints; each one is a known failure mode.

A plan also fails if it assumes a single setup for a part that cannot physically be reached in one. Five-axis access is generous but not unlimited. Undercuts, internal corners and deep bores still need a second op, and the plan should say so plainly.

Finally, a plan that ignores quantity is incomplete. One prototype and a 10,000-part run are different documents. The prototype plan optimizes for speed and fixture simplicity. The production plan optimizes for repeatability and tool life. We quote both from one drawing, with no minimum order quantity.