CNC Tour Operating Procedures: How a Machining Cycle Is Run

• 1
  What CNC tour operating procedures actually cover
• 2
  Pre-start checks before the first chip
• 3
  How CNC tour operating procedures set feed, speed, and depth
• 4
  In-process monitoring: what the operator watches and records
• 5
  Shutdown, chip clearing, and handover
• 6
  When a standard tour procedure is not enough
• 7
  Tour phase, what is checked, and the failure it prevents

A tour is one complete run of the machine through a part: power-up, reference return, load, cut, unload, shutdown. The procedures around that run are the written rules for each transition. They exist because most scrap is not caused by a bad program. It is caused by a chip left on a locating face, a tool pulled short in the holder, or a warm machine referenced cold.

On a vertical mill the cycle is easy to watch. On a horizontal boring mill with a Ø400 mm rotary table the same steps take longer and the failure modes change, because the pallet and the boring bar both have to be proven before the first cut. The procedures are different by machine class, but the logic is the same: prove the reference, prove the tool, prove the part, then cut.

For buyers, this matters at the quotation stage. A shop that can describe its tour procedures in specific numbers is a shop that can hold ±0.005 mm across a 10,000 part run. A shop that answers with adjectives cannot. The rest of this page breaks the cycle into the checks that carry the tolerance.

The machine is powered up and returned to reference before anything is loaded. Axis reference return on a cold machine is not the same as on a warm one. A casting that was 18 °C overnight can move 0.02–0.04 mm over the first hour of spindle running, which is more than the tolerance on many bores. So the reference is taken, then the machine is allowed to idle or run a warm-up cycle until the bed temperature is stable.

Air and coolant come next. Coolant concentration is checked with a refractometer; a typical range is 6–10% for aluminum and 8–12% for steels and stainless. Too lean and tools wear fast, too rich and the sump foams. Way lube levels and air pressure are checked at the same time, because a pressure drop mid-cut will trip an alarm and leave a tool in the part.

The workholding is proven before the part is loaded. Vise jaws are indicated, chuck jaws are checked for bell-mouth, and fixture clamps are torqued to the drawing value. A clamp that is 10% loose will let the part lift on a heavy facing pass, and the symptom shows up as a taper that looks like a machine error.

• 1
  Reference return on a warm machineIdle or warm-up cycle until bed temperature is stable, then reference.
• 2
  Coolant at 6–10% for aluminumRefractometer reading, not a visual check.
• 3
  Fixture torque to drawing valueLoose clamping shows up as taper, not as chatter.

Cutting data is not guessed at the control. It is written into the setup sheet from the tool library, then adjusted at the machine within a narrow band. A typical 12 mm carbide end mill in 6061-T6 runs at 2,500–3,500 rpm with a surface speed around 300–500 m/min, a feed of 0.05–0.10 mm per tooth, and an axial depth of 0.5–1.0 × diameter in roughing. In 17-4PH stainless the same tool drops to 60–90 m/min surface speed, and depth of cut comes down to 0.2–0.4 × diameter.

The first cut is a sensor reading, not a production cut. The operator listens for a change in pitch, watches chip color and shape, and reads the spindle load meter. Aluminum should throw bright, well-formed chips and pull 40–60% of rated load. Steel chips that come off brown or blue mean the surface speed is too high or the coolant is not reaching the edge. A load meter that climbs through the pass means a tool is dulling or a chip is packing.

Depth and stepover are the two dials that decide whether the part stays in tolerance. Heavy radial cuts on a thin wall will push the wall away from the cutter and spring back after the pass, so the bore measures undersize and the wall measures oversize. On thin-walled aluminum parts we drop radial engagement to 5–10% of cutter diameter and raise axial depth instead. The cut is quieter and the wall stays where the model says it should.

Roughing and finishing are separate operations with separate tools. Roughing leaves 0.2–0.5 mm of stock on finish faces. Finishing tools are kept for finishing only, because a tool that has roughed 4140 will not hold Ra 0.8–1.6 μm on a bearing seat.

• 1
  Read the load meter on the first cut40–60% of rated load in aluminum, lower in stainless.
• 2
  Reduce radial engagement on thin walls5–10% of cutter diameter, deeper axial passes.
• 3
  Keep finishing tools separateA roughed tool will not hold a fine finish.

Once the cycle is proven, the job becomes a monitoring task. The operator checks the first part fully, then spot-checks at a set interval. On a run of 200 aluminum housings we check the first part, then every 10th part on the critical bore, and every 20th on the outer profile. If a dimension is drifting toward a limit, the interval shortens. The interval is part of the procedure, not a decision made on the day.

Tool wear is tracked by cut count, not by calendar. A 6 mm carbide drill in 6061 may hold diameter for 800–1,500 holes; the same drill in 316L may hold for 150–300. The tool change point is set at 80% of the observed life so the last 20% of the run is not spent making parts at the edge of the tolerance band. Offsets are adjusted at the control, and the adjustment is logged.

Temperature drift is the quiet one. A machine that ran a long roughing cycle will have a warmer column and a slightly different Z position than it did at start-up. On parts with a 0.01 mm true-position callout, the operator re-references or re-checks the first feature after any long heavy cut. This is the difference between a process that holds ±0.005 mm and one that holds it only on the first part.

The end of a tour is its own procedure. The spindle stops, the coolant is left running for 30–60 seconds to wash chips off the way covers, and the table is moved to a park position that keeps the covers relaxed. Chips left on a locating face overnight will be pressed into the surface on the next load. Chips left in a T-slot will sit against a clamp and change the clamp height by a few hundredths.

Aluminum and steel parts are separated before shutdown because wet aluminum chips and steel chips in the same bin create a corrosion cell. Parts that will sit overnight get a light oil film. Coolant is left circulating on a timer in some shops to stop bacterial growth; the refractometer reading is recorded for the next shift.

The handover is written, not spoken. The next operator needs to know the tool that is due for change, the offset that was adjusted, the dimension that was drifting, and the part count in the run. A twenty-second note at the machine saves a first-part scrap on the next shift, which is the most expensive scrap of the day.

Standard procedures assume a stable part and a stable machine. Some jobs break both assumptions. A 4,000 mm long extrusion on a gantry mill will change shape as the material is removed, because internal stress is released. The procedure has to add a stress-relief pass, a re-clamp, and a re-reference between roughing and finishing. Cutting the whole part in one setup will produce a banana, not a straight rail.

Thin floors and tall ribs behave the same way. A 1.5 mm floor on an aluminum electronics housing will deflect under the cutter and spring back, so the measured floor is thicker than the programmed one. The procedure adds a semi-finish pass at low load and a final pass with a small stepover. On a part like this, more passes at lower load are faster than fewer heavy passes, because the heavy pass has to be scrapped and rerun.

Five-axis work adds a rotary axis to the reference chain. The trunnion has to be centered and the rotary zero proven with a test bar before the first part, and the rotary position is re-checked after any crash or alarm. On a mill-turn center the sub-spindle adds a second reference, and the transfer position has to be verified on the first part of every run. These checks add time, but they are the reason a 16-station 5-axis cell can run unattended overnight and still hold the print.