Consistent CNC Machining Services: How to Check a Supplier Before You Order

• 1
  What separates a consistent shop from a lucky one
• 2
  What to check, what to ask, what a weak answer looks like
• 3
  Tolerance repeatability: the number that actually matters
• 4
  Documented process control and setup discipline
• 5
  Inspection, materials and finish: where batch drift shows up
• 6
  Lead time, order size and the quote you can compare
• 7
  How to vet a supplier in seven steps

Consistency starts with repeatability. A supplier who holds ±0.005 mm across a 500-piece run is more useful than one who hits ±0.002 mm on a single sample and drifts after the first tool change. Ask how tolerance is verified on part 1, part 50 and part 500 of the same order. If the answer is only "we inspect the first article," the process is not controlled yet.

Thermal drift is the usual culprit. A spindle that runs 8 hours straight grows; coolant temperature moves the workpiece; chip load changes as a tool wears. Shops that control this log coolant temperature, set tool wear limits in the CAM program, and re-measure critical features on a fixed interval rather than a fixed feeling.

For most industrial parts, ±0.005 mm is enough. Tightening beyond that raises cost fast because it forces slower feeds, more in-process checks and scrap risk. Decide the tolerance from function, then check that the shop can hold it repeatedly. Printing ±0.002 mm on a drawing that does not need it just moves the risk to your supplier and back to your assembly line.

A useful test: send a part with one critical bore and several loose features. Watch whether the shop spends its attention where the tolerance is tight, or treats every dimension the same. That choice tells you how they will treat your production order.

Consistency lives in paperwork as much as in metal. Every step from material receiving to packaging should follow a written procedure: who sets the machine, which fixture, which offsets, which inspection points. When a technician is out sick or a job moves to a second machine, the output should not change.

Ask to see a setup sheet for a part similar to yours. A good one lists fixture, work offset, tool list with wear limits, and the first-article dimensions to confirm before running production. A weak one says "program as per drawing." The setup sheet is where repeatability is either designed in or lost.

Machine swapping is the hidden risk in multi-plant operations. If your parts can run on more than one machine or in more than one plant, the process must be portable: same fixture concept, same tool strategy, same inspection plan. Otherwise batch two will not match batch one, and you will find out at assembly.

GreatLight runs three wholly-owned plants with 127 high-precision CNC machines, including 16 simultaneous 5-axis centers and 16 mill-turn centers. The same setup documentation travels with the job across plants, which is how a 4,000 mm maximum processing size and a Ø400 mm rotary table job can both follow one control plan.

Inspection should be built into the run, not bolted on at the end. Raw material check on arrival, in-process monitoring at defined intervals, and a final inspection before shipment. Reports are available on request. If a shop can only show you a final inspection stamp, you have no way to know when a drift started.

Material traceability matters more than most buyers expect. A mill certificate tied to a heat number lets you confirm that batch three used the same 6061-T6 or 17-4PH as batch one. Without it, a substitution can change machinability, hardness after heat treatment, and the surface finish you get from the same program.

Surface finish is the most common silent drift. Ra 0.8–1.6 μm is a normal as-machined target for many parts, while Ra 0.2–0.8 μm needs slower passes and tighter tool control. If the shop quotes a finish by name instead of a measured Ra range, expect variation between batches, especially after a tool change or a switch of material supplier.

Post-processing adds another variable. Anodizing, electroless nickel, black oxide and bead blasting each change dimensions slightly. A consistent supplier documents the pre-finish allowance and checks the part after coating, not before. That is the difference between a batch that fits and a batch that needs rework.

Lead time claims are easy to make and hard to verify. Ask what happens between purchase order and shipment: when the material is ordered, when the first article is approved, when production starts. A shop that can quote and return a free DFM analysis within 12 hours and start production within 24 hours is usually a shop with real capacity, not a broker.

Order size is a consistency question. A supplier willing to run one prototype and a 10,000+ part run on the same control plan is telling you something useful: the process is standard, not improvised for large jobs. No minimum order quantity removes the pressure to inflate a small order into a big one.

Compare quotes on the same basis. Ask for the split between material, setup, cycle time and finishing. A single lump-sum number cannot be compared, and it hides where the cost will move when your design changes. Cycle time per part and setup hours per order are the two numbers that tell you whether volume pricing is real.

Certifications belong in this comparison. ISO 9001:2015 covers general quality systems. IATF 16949:2016 is what automotive and EV programs expect. ISO 13485:2016 applies to medical device components. ISO 27001:2022 covers information security, which matters when you upload CAD files. Check the scope on each certificate, not just the logo.