CNC Expo 2024: What the Show Floor Signals for Part Buyers

• 1
  Why five-axis kinematics dominated CNC Expo 2024
• 2
  In-process probing: what it corrects and what it cannot
• 3
  Shop-floor data: what is worth logging, and what is noise
• 4
  Hard materials and the tooling they force
• 5
  Where these trends do not apply to your part
• 6
  Matching the process to the part

A three-axis machine moves the tool in X, Y and Z while the part stays still. A five-axis machine adds two rotary axes, so the tool can tilt and the table can rotate under it. That extra motion does one real job: it keeps the tool normal to the surface and lets one setup reach five faces of a part. The 2024 show floor made this concrete rather than theoretical. Simultaneous motion lets a single fixturing reach features that used to need three separate operations.

The mechanism matters because every refixturing step adds stack-up error. Clamp a part, cut it, unclamp it, turn it, clamp it again, and the second datum rarely lands exactly where the first one did. Each move adds a few microns of positional drift. Five-axis work removes most of those moves. A part that once needed four setups may need one. Fewer setups means a shorter tolerance chain, and that is the engineering payoff, not the axis count itself.

There is a boundary. Five-axis does not fix a part that is flimsy or badly fixtured. Thin walls still deflect under cutting force no matter how the tool approaches. Deep pockets with a small tool still chatter if the tool overhang is long. When we quote a part, we look at wall thickness, depth-to-diameter ratio and whether the geometry actually requires tool tilt. If a three-axis setup reaches every face within tolerance, adding rotary axes only raises the hourly rate.

A touch probe mounted in the spindle measures the part between cuts. The control compares the result to the model, then shifts the work offset or re-cuts the feature. This closes the loop inside the machine instead of waiting for a coordinate measuring machine hours later. The benefit is real for castings and forgings, where stock varies from part to part. Probing finds the actual surface and re-datums the toolpath to it.

What it cannot do is rescue a process that is already out of control. A probe measures position, not surface finish, not subsurface cracks, not residual stress. If a boring bar is walking because the insert is worn, probing will keep correcting the offset while the hole grows oval. The correction hides the symptom. That is why probe data has to be logged, not just consumed. Trend a feature over 50 parts and the drift shows up before the scrapped parts do.

The engineering meaning for a buyer is about which features get probed. A critical bore with a ±0.005 mm tolerance is worth probing. A clearance slot at ±0.2 mm is not. Ask a supplier which features are probed and how often the probe is recalibrated against a known artifact. A probe that drifts is worse than no probe, because it moves the process in a confident wrong direction.

Thermal growth is the other limit. A spindle running for four hours grows in Z. Probing catches the resulting error only if it runs often enough to track the drift. On a long run, probe the first part, then every tenth, then compare. If the offsets trend in one direction, the machine is warming, not the part moving.

Every modern control emits data. Spindle load, axis current, tool life counters, cycle time. The temptation is to log all of it and call it a smart factory. The useful subset is smaller. Three signals change decisions: tool wear, spindle load against a baseline, and the offset corrections the probe makes. Those three tell you when a process is drifting and which tool is causing it.

Spindle load is the one people underuse. A roughing end mill cutting 6061 aluminum draws a steady current. When that current climbs 15 percent on the same program with the same material, the tool is dulling or the chip evacuation is failing. Catch it and you change the insert on schedule. Ignore it and you get a rough surface and a scrapped part near the end of the run.

Data does not replace inspection. A 100 percent inspection step before shipment still catches what the sensors miss, including handling damage and burrs. The sensors tell you the process is stable. Inspection tells you the part is right. They answer different questions, and a shop that confuses them will ship a statistically stable batch of wrong parts.

The cost side matters too. Adding sensors, a data historian and someone to read the dashboards is real money. For a 50-part prototype run, that overhead rarely pays back. For a 10,000-part run in Inconel, where a single tool failure costs a scrapped part worth hundreds of dollars, it pays back in weeks. Match the instrumentation to the run length.

The 2024 exhibits leaned hard into titanium, Inconel and 17-4PH stainless. These materials change the whole cutting equation. Titanium Ti-6Al-4V conducts heat poorly, so the heat stays in the cutting edge instead of leaving with the chip. Tool life drops, and the surface can work-harden if the feed is too light. You cut titanium with a heavy feed and a moderate speed, not the other way around.

Inconel is worse for tool wear and better for heat resistance, which is exactly why it is used in hot sections. It work-hardens fast, so a rubbing pass is a ruined surface. The rule on the floor is simple: never let the tool dwell. Keep the feed high enough that the edge bites under the hardened layer rather than skating across it.

17-4PH sits between the two. In the solution-treated condition it machines like a 300-series stainless. After aging it hardens and the same cut becomes difficult. If a print calls out aged 17-4PH, the machining sequence has to be planned around the heat treat, not bolted on afterward. Cut first, age second, then finish only what the heat treat moved.

Magnesium AZ31B and AZ91D came up too, mostly in lightweight housings. These cut fast and finish well, but the chips are flammable. The process controls are about chip management and coolant choice, not spindle speed. A supplier who has not run magnesium before will learn this the hard way.

A simple prismatic bracket with three drilled holes and two milled faces does not benefit from any of this. A three-axis machine with a good vise holds ±0.05 mm all day. Adding five-axis motion, probing and a data historian to that part raises the price and changes nothing about the result. The right process is the cheapest one that holds the tolerance.

Prototype quantities are the second boundary. On a five-part run, setup dominates the cycle time. Probing each part adds minutes that never come back. For one-off parts, a coordinate measuring machine inspection after the fact is more useful than in-process probing, because you want the full dimensional report, not a work-offset correction.

The third boundary is geometry that no rotary axis can reach. A deep, narrow slot with a square internal corner still needs EDM or a broach. Five-axis tilts the tool, but a rotating cutter is round. It cannot leave a sharp internal corner without a separate operation. When a print shows a true square corner, the process plan includes something other than milling, and the quote should say so.

GreatLight runs 16 simultaneous 5-axis centers alongside 27 three-axis machines. That mix exists because most parts do not need the rotary axes. Putting a job on the right machine is a bigger cost lever than putting every job on the newest one.