What to Look For in a CNC Machine

• 1
  What to Look For in a CNC Machine: Axis Count
• 2
  Travels, Table Size and Part Mass
• 3
  Spindle Speed, Torque and Taper
• 4
  How Accuracy Is Actually Proven
• 5
  Fixtures and Workholding Decide the Result
• 6
  Machine Class vs Part Type

A 3-axis mill moves the tool in X, Y and Z while the part stays still. The setup is simple, the machine is stiff, and for flat plates, pockets, and drilled hole patterns it is the cheapest way to hold tolerance. The limit is not accuracy, it is access. Any feature on the side walls or the far face needs a second setup.

A 4-axis machine adds a rotary table, usually turning around X. Now you can index to four faces without unclamping. This is the standard answer for parts with holes on multiple sides, keyways, or a bolt circle on an end face. Add a tailstock and the same machine turns round stock as well.

A 5-axis machine moves the tool or the table on two extra rotary axes at the same time. The gain is not speed. It is access. A ball nose cutter can stay normal to a curved surface for the whole pass, so deep pockets and undercut flanks get cut in one setup.

Angle errors multiply over long tools. On a 5-axis machine with a Ø400 mm rotary table, a 0.002° indexing error at the table edge is small. The same error on a 300 mm tool extension is not. That is why the rotary axis, not the linear axis, usually sets the floor on a complex part.

Published travels are measured at the tool tip with no fixture in the way. A machine listed at 750 × 1,150 × 550 mm loses a large slice of that to the vise, the chuck, and the tool holder. Always ask for the usable envelope with a standard fixture installed.

Part mass matters as much as part size. A 500 kg block on a 600 × 600 × 600 mm machine is fine. The same block on a compact 500 × 310 × 200 mm machine will deflect the table and wear the ways. Match mass to the machine class, not just to the drawing.

Long parts need a different check. Our largest travel is 4,000 × 400 × 150 mm, which suits long extrusions, rails, and structural profiles. Thin and long is the hard case: the part moves more than the machine does. Support it every 300–400 mm or the middle will chatter.

When the part is bigger than any single machine, the answer is often two setups and a dowel-pin datum, not a bigger machine. Ask how the shop holds the second-setup alignment. If the answer is a probe and a known datum, that is a real process.

Spindle speed and torque trade against each other. A 20,000 rpm spindle with an HSK or BT30 taper is built for 3–8 mm cutters in aluminium, finishing at Ra 0.8–1.6 μm. It will stall in a 50 mm face mill in 4140 steel. A 6,000 rpm, 40-taper spindle is the opposite.

Read the torque curve, not the peak kW number. Peak power is often quoted at a speed you will never use. What matters is the continuous torque band where your cutter actually runs. If the curve is not published, ask for it.

Tool holding is part of the spindle. Runout at the tool tip should stay within 0.005 mm for finishing work. A worn holder or a chip under the taper will show up as a size shift between the roughing and finishing pass.

On titanium and Inconel the limit is heat, not force. TC4 (Ti-6Al-4V) conducts heat poorly, so most of it goes into the edge. Lower surface speed, higher feed per tooth, and flood or through-tool coolant. A high-speed spindle alone will not fix a titanium cut.

The nameplate says ±0.005 mm. That number is meaningless without a measurement chain behind it. Positions repeat. Thermal growth does not. A spindle that runs 8 hours at 12,000 rpm can grow several hundredths of a millimeter in Z.

Ask how the shop compensates. Warm-up cycles, spindle growth sensors, and in-process probing are the normal answers. A machine that is probed and re-datumed between operations holds size far better than one that is trusted for a full shift.

Then look at the inspection side. A part is only verified if it is measured on equipment with its own calibration record. Coordinate measuring machines, optical comparators, and laser scanners cover different feature types. A CMM checks a hole position. It does not check a thread form.

Finally, look at the sampling rate. 100% inspection before shipment is a different claim from AQL sampling. For medical and aerospace work, ask which one applies and what report you get. We run raw material checks, in-process monitoring, and final inspection, with reports on request.

A rigid machine with a weak fixture behaves like a weak machine. The first mode of vibration usually lives in the workholding, not the spindle. Soft jaws bored on the machine, a proper vise, or a dedicated plate all beat a generic clamp setup.

Thin-wall parts are the classic case. A 1.5 mm wall in aluminium will move under jaw pressure even at low clamping force. The fix is often to rough, stress-relieve, then finish with light passes and low clamping load.

For 5-axis work the fixture must allow tool access to every face. That usually means a dovetail block or a self-centering vise on a tombstone, not a stack of parallels. Ask to see the setup photo before the job runs.

Zero-point systems cut setup time and improve repeatability across runs. If a shop runs the same family of parts every month, a pallet system pays for itself in size consistency, not just in hours.