Precision CNC Wholesale: How Repeat Runs Really Work

• 1
  Why a Wholesale Order Is a Different Problem
• 2
  Where ±0.005 mm Comes From in a Repeat Run
• 3
  Fixtures Decide Whether the Lot Stays Inside Tolerance
• 4
  Material Lots, Heat Treat and the Hidden Variable
• 5
  Inspection Plans That Scale Past the First Article
• 6
  When Batch Buying Pays Off and When It Does Not

A single prototype proves that a geometry can be cut. A precision cnc wholesale order proves that the same geometry can be cut in March as accurately as it was cut in the first week. Those are two different engineering problems, and the second one is where most schedules slip.

When a shop quotes a batch, it is not only pricing spindle time. It is pricing the fixture, the tool life plan, the in-process gauging and the paperwork that keeps part 4,000 interchangeable with part 1. That is why a wholesale unit price can look high next to a one-off quote and still be the cheaper route.

The mechanism is straightforward. Variation comes from four sources: the machine, the setup, the material lot and the operator. On a single part you can chase each source by hand. On 5,000 parts you have to design them out, because nobody can inspect their way to consistency at that volume.

Order size changes which of those four sources dominates. At 50 pieces, setup error is the main risk. At 5,000 pieces, tool wear and thermal drift take over. The control plan has to follow the volume, not the drawing alone.

A ±0.005 mm tolerance is a process capability statement, not a wish. It means the shop has measured the spread of its own process on that feature and knows the spread fits inside the band with room left over. Without that measurement, the number on the drawing is a gamble.

Repeatability is the part that batch buyers underestimate. A machine that hits ±0.005 mm once and ±0.02 mm the next morning is worse than a machine that holds ±0.01 mm all week. Wholesale pricing depends on the second behavior, because it removes the need for sorting and rework.

Practical rule: keep the tight band on the features that mate, seal or locate. Everything else can sit at ±0.05 mm or looser. When every dimension on a print carries the same tight tolerance, the shop has to fixture and gauge for the worst one across the whole part, and the price reflects that.

Surface finish follows the same logic. Ra 0.8–1.6 μm is a normal machined finish for mating faces. Ra 0.2–0.8 μm needs a finer step-over, a different tool and often a second operation. Applying it to a non-contact surface adds cost without adding function.

In a repeat run, the fixture is the product. The cutting program can be perfect and still produce drift if the blank locates differently on part 900 than it did on part 9. Soft jaws machined in place, dedicated nests or a hydraulic tombstone all solve this, and the choice depends on how many sides you need to reach.

For prismatic parts, a machined soft jaw with a hard stop gives you a repeatable Z datum and lets the operator load blind. For parts with an organic outer profile, a cast or 3D printed nest plus vacuum or low-pressure clamping usually beats vise work, especially when the wall is thin.

Five-axis work changes the fixture calculus again. A Ø400 mm rotary table with a tombstone lets you cut four or five faces in one setup, which removes the re-datum error that creeps in when a part is moved between operations. Fewer setups means fewer chances to lose the tolerance band.

The failure mode to watch for is fixture wear. Aluminum soft jaws lose their edge after a few hundred cycles. If the shop does not re-cut or replace them on a schedule, the first 200 parts pass and the next 300 drift. Ask what the replacement interval is.

Two bars of 6061-T6 from different heats can machine differently. Hardness varies, residual stress varies, and a part that held flat on Monday may bow after the same cut on Thursday. For long runs, buy the whole lot up front and keep it under one heat number.

Residual stress is the quiet one. A part hogged out of stressed plate will move after it is unclamped, and it will keep moving after anodizing. Stress-relief before finishing, or a rough-machine and re-clamp cycle, removes most of that movement before the tolerance band is measured.

Heat-treated steels behave the same way but with a larger swing. If a print calls for 4140 at 28–32 HRC, the shop needs the heat treat lot tied to the machining lot. Mixing them means the second batch may need a different speed and feed recipe to hit the same finish.

For titanium and Inconel, tool wear is the dominant variable rather than material variation. TC4 (Ti-6Al-4V) and Inconel work-harden at the cut, so a worn insert starts rubbing instead of shearing. Tool change intervals have to be counted in minutes of cut time, not in parts.

First article inspection proves the setup. It does not prove the lot. A wholesale inspection plan needs three layers: raw material verification, in-process checks at set intervals, and a final check before shipment. Skip the middle layer and any drift is found at the end, when the parts are already finished.

The interval matters more than the method. Checking one part every 20 cycles catches a slow drift. Checking one part per shift catches a fast one. Which interval you need depends on the tolerance band and the tool life, and it should be written into the process sheet rather than decided by the operator.

CMM reports are useful, but they are a snapshot of the parts measured, not a guarantee for the rest. For critical features, gauge the feature directly at the machine during the run and keep the CMM for the final audit. That way the correction happens before the parts are off-spec.

GreatLight inspects 100% of parts before shipment, with raw material checks, in-process monitoring and a final check. Reports go out on request. For medical and automotive work, the inspection records are tied to the lot so a customer can trace a shipment back to the material heat.