Secondary CNC machining: key steps for holding tight tolerances

• 1
  Key takeaways
• 2
  Why secondary CNC machining exists
• 3
  Read the first operation before you fixture anything
• 4
  Workholding choices that protect a finished surface
• 5
  What goes wrong on the shop floor
• 6
  Secondary CNC machining key steps, in order
• 7
  When a second setup helps, and when it does not

A part comes off the first operation with most of its geometry done. Then something is still wrong. Two holes sit 0.03 mm off true position. A bearing bore needs Ra 0.4 μm instead of the Ra 1.6 μm the roughing pass left behind. A customer revised a pocket depth after the blanks were already cut. Secondary CNC machining is how those gaps close without scrapping the part.

The reasons repeat across shops. Multi-face parts that cannot be reached in one setup. Tolerances that the first op could not hold because the part was too thin and deflected. Features added late in the design cycle. Legacy parts that need repair inside the original material envelope. Hardened tool steel that only becomes machinable after a finish pass with the right insert.

There is a cost line here that engineers should watch. A second setup adds fixturing, re-datuming, and an extra inspection loop. On a 10,000-part run that is real money. It is still cheaper than a scrap rate above 1%, which is what happens when you try to force everything into one operation. The judgment call is whether the tolerance you need is reachable in setup one at all.

• 1
  ReachFaces and holes the first setup cannot present to the spindle.
• 2
  ToleranceDeflection or residual stress pushed the first op out of spec.
• 3
  RevisionThe drawing changed after blanks were cut.
• 4
  FunctionA bore, seat, or thread that must be sized to a mating part.

Before a second setup is programmed, the finished surfaces from setup one need measuring. Not the nominal values on the drawing. The actual values on the parts in front of you. If the first op ran 0.02 mm wide on a critical width, a fixture built to nominal will clamp the part off-center and the error carries straight into the second operation.

We measure three to five parts from the first lot, log the spread, and use the mid-point as the programming reference. On a typical ±0.005 mm job that 0.02 mm shift is four times the total tolerance. It cannot be ignored or averaged away later.

This is also where you check for residual stress. A part that was hogged out of 7075 or 17-4PH can move 0.05–0.15 mm in the hours after the first cut. If the second setup runs too soon, the part moves again after the fixture is released and the final inspection fails. Stress relief, or a dwell between operations, is the fix.

• 1
  Measure the lot, not the drawingUse real first-op values as the programming reference.
• 2
  Log the spreadA 0.02 mm lot spread eats 4× a ±0.005 mm tolerance.
• 3
  Watch for movement7075 and 17-4PH can shift 0.05–0.15 mm after roughing.

The second setup usually clamps on a surface that already meets a tolerance. That surface cannot be damaged. Soft jaws machined in place to the actual part profile distribute load better than hard jaws and are the default for aluminum and stainless parts under 150 mm. Cut them 0.05 mm under the measured part size so the part seats without rocking.

For thin walls, clamping force is the enemy. A 2 mm aluminum wall gripped at 1.5 kN will deflect about 0.03 mm, and the cut happens in that deflected position. The part springs back when the vise opens and the wall ends up thin. Vacuum chucks or low-pressure hydraulic vises under 0.8 kN hold the part without that distortion.

Odd geometry needs dedicated fixtures. A machined pocket that matches the finished contour, plus a single clamp over a non-critical pad, beats three clamps chasing flatness. On our 5-axis centers with a Ø400 mm rotary table, we often cut the fixture and the part in the same program so the fixture profile is correct by construction.

• 1
  Soft jawsMachined in place, 0.05 mm under measured part size.
• 2
  Low clamp forceStay under 0.8 kN on walls thinner than 3 mm.
• 3
  Dedicated nestFor irregular parts, one pocket and one clamp beat three.

The most common failure is using a rough surface as the second datum. The first operation left a saw-cut or cast face that varies across the lot, and the second setup chases that variation. Every hole drilled from it inherits the error. Fix the datum, fix the part.

Second is over-clamping. Engineers specify a fixture, the operator tightens it until the part stops moving, and a thin wall deflects. The part measures correctly while clamped and fails after release. The fix is a torque spec on the vise, not a bigger clamp.

Third is ignoring heat. A finish pass at high speed raises the part temperature by 5–15 °C. Measure immediately and the reading is wrong by 0.01–0.02 mm on aluminum. Let the part cool to room temperature, or measure with a known temperature correction, before you sign off the inspection report.

Fourth is running a finish tool on a roughing allowance. A 0.05 mm radial cut with a tool ground for 0.3 mm will rub, work-harden the surface, and produce a finish worse than the roughing pass. Match the tool to the allowance, or leave the right allowance for the tool.

• 1
  Rough datumLot variation of 0.1 mm or more carries into every dimension.
• 2
  Over-clampingPart measures fine clamped, fails after release.
• 3
  Hot measurement5–15 °C part temperature reads 0.01–0.02 mm off on aluminum.
• 4
  Tool-allowance mismatchRubbing work-hardens the surface and ruins the finish.