CNC Machining Batch Customized: How Repeat Runs Stay Identical
A batch is not a big prototype. Once you cut the same part 50 or 500 times, setup, tool wear, fixture clamping and inspection frequency decide whether run 2 matches run 1. This page explains the mechanics, the limits, and when a batch route is the wrong choice.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
Why a CNC machining batch customized order behaves differently from a prototype
A prototype is a single event. A CNC machining batch customized order is a controlled repetition of that event, and repetition is where variation enters. The first part tells you the program works. Parts 2 through 500 tell you whether the process holds. Between them sits everything that drifts: cutter flank wear, thermal growth in the spindle and ballscrews, chip buildup on the fixture, and the slow relaxation of clamps on a part that was straight when it was loaded.
The core mechanism is that a CNC machine does not reproduce a shape. It reproduces a toolpath. If the tool has worn 0.02 mm on the flank since the last touch-off, every following part is cut 0.02 mm smaller than the one before it. Nothing in the control knows this. The offset was correct at the start of the run and stays correct in the controller's memory until a person measures a part and changes it.
That is the whole difference. Prototype tolerance is a programming and setup problem. Batch tolerance is a drift problem. You manage it with tool life limits, scheduled offset changes, and inspection that happens often enough to catch the drift before it reaches the tolerance band, not after.
On a 200-piece run held to ±0.05 mm, drift is rarely the limit. On the same run held to ±0.005 mm, a single worn 6 mm end mill can consume the entire band in under 100 parts. Material matters too: aluminium 6061 and 7075 cut clean and hold size, while 316L stainless and Ti-6Al-4V work-harden, push cutting forces up, and wear tools two to three times faster.
- 1Prototype thinkingOne good part proves the program
- 2Batch thinkingEvery part must land inside the band
- 3The variableTool wear and thermal drift, not the code
Fixtures and workholding: the hidden variable in repeat runs
A vise is fine for ten parts. For five hundred, it becomes the largest single source of scatter. Every load is a new clamp position, and manual clamping force varies between operators and between shifts. On a thin-walled aluminium housing, one extra quarter turn on the handle can bow the wall 0.03 mm, which is then machined into the part and springs back when the clamp releases.
The fix is a dedicated fixture that locates on a machined datum rather than on a raw surface. Once the fixture is dialed in, the operator loads the part against hard stops and clamps to a torque value or a pneumatic pressure, not by feel. Repeatability then comes from the fixture, and the operator only has to confirm the part is seated.
For our batch work we build soft jaws, vacuum plates, or tombstone fixtures depending on the geometry. A tombstone on a 5-axis machine with a Ø400 mm rotary table lets several parts be cut per cycle, which removes a load-unload cycle for each part and cuts both labor and clamp variation.
Fixtures cost money up front. On a 50-piece run, a fixture that takes eight hours to design and build may not pay back. On a 500-piece run it usually does, because the alternative is scrap and rework. That payback point is one of the first things we check when quoting a batch.
- 1Vise onlyAcceptable up to roughly 20–30 parts
- 2Soft jawsGood for prismatic parts with a clean datum
- 3Tombstone or vacuum plateFor thin walls and 5-axis multi-part cycles
Holding ±0.005 mm across 500 parts: what actually has to be controlled
Tolerance is a system number, not a machine number. A machine that can position to ±0.005 mm on a cold morning will not hold that through a six-hour run unless the shop controls the things around it. The first is temperature. A 1 °C change in a 100 mm aluminium part moves it about 2.3 μm. A shop that swings 6 °C between morning and afternoon moves the part 14 μm without touching the control.
The second is the measurement itself. You cannot hold ±0.005 mm with calipers. Batch work at this level uses micrometers, bore gauges, and a CMM, and the gauges are checked against a reference before the run. Our inspection flow starts with a raw material check, moves to in-process monitoring at defined part counts, and ends with a final inspection before shipment. Reports are available on request.
The third is the offset strategy. Instead of letting a tool run until it visibly fails, we set a tool life count based on test cuts, then stop and re-measure at that count, or change the tool on a schedule. That converts a random drift into a planned step.
This is why qualification rate matters more than peak accuracy. We run at a 99.99% qualification rate, which means the process is designed so that parts do not wander out of band in the first place. Chasing a 0.005 mm band with constant rework is not a process. It is luck with paperwork.
- 1TemperatureControl the room before you control the offset
- 2MetrologyMicrometers and CMM, not calipers
- 3Tool lifeChange on a count, not on a hunch
When CNC machining beats casting or molding for a customized batch
CNC is the right route when the geometry is still moving, when the volume is under a few thousand parts, or when the material is one that casting cannot match. Tool steel, 17-4PH, Inconel and titanium are common examples. A die for these materials costs more than the parts, and it locks the design before the design is finished.
CNC also wins when the part has features that a mold would struggle with. Undercuts, deep pockets with tight corner radii, threaded ports, and sealing faces that need Ra 0.8–1.6 μm are all routine on a mill and difficult or expensive in a mold. Free DFM analysis within 12 hours is where most of these trade-offs get caught.
Where CNC loses is at high volume with simple geometry. A 50,000-piece bracket in ADC12 should be die cast, not milled. A 100,000-piece housing in ABS should be injection molded. If a customer asks us to mill those, we say so, because the per-part cost will not survive the comparison and the lead time will not either.
The middle zone is real, though. Between roughly 100 and 5,000 parts, CNC often wins on total cost once you count tooling, lead time, and the probability that the design changes. That is the zone where a CNC machining batch customized order makes the most sense.
- 1Choose CNCDesign still moving, hard material, tight features
- 2Choose casting50,000+ simple parts in a castable alloy
- 3Choose molding100,000+ plastic parts with frozen geometry
Material behavior across a batch: what changes part to part
Material is not a constant. Two bars of 6061-T6 from different heats can machine differently, and the difference shows up as chip formation and surface finish rather than as a dimensional error. We buy to spec and check incoming stock, but the practical effect is that the first article of a new heat is treated as a fresh setup, not a continuation.
Stainless tells the story more clearly. 303 machines freely with good chip control, which is why it is the default for small batch turned parts. 316L and 17-4PH work-harden, so a dull tool rubs instead of cutting, the surface tears, and the next part may need a different offset. On a 500-piece 316L run, tool changes are planned, not reactive.
Aluminium 7075 holds size well and takes a fine finish, but it is more notch-sensitive than 6061 and can move when a lot of material is removed from one side. Titanium TC4 (Ti-6Al-4V) and Inconel need low surface speed, rigid setups and flood coolant; they are the materials where a batch run most depends on the fixture being right.
Plastics are their own problem. POM and PEEK move with temperature and moisture, and clamping force that is fine on steel will deform them. For these parts, we cut with sharp single-flute tooling, use light finishing passes and control the coolant, because heat is what moves the part between operations.
- 1Aluminium6061, 7075 and 6082 hold size well
- 2Stainless303 free-cutting; 316L and 17-4PH need tool life control
- 3Titanium and InconelLow speed, rigid fixture, flood coolant
- 4PlasticsSharp tooling, light passes, temperature control
Batch route and process choice by volume and geometry
Use this as a first filter. It is a cost and risk comparison, not a rule book.
| Volume | Typical route | Main limit | Watch for |
|---|---|---|---|
| 1–20 parts | 3-axis or 4-axis mill, vise | Setup dominates cost | Manual clamp variation |
| 20–200 parts | CNC with soft jaws | Tool wear drift | Offset checks at set counts |
| 200–2,000 parts | CNC with dedicated fixture | Fixture payback | First-article re-approval per heat |
| 2,000–10,000 parts | Mill-turn or 5-axis multi-part | Cycle time per part | Chip evacuation, thermal growth |
| 10,000+ simple castable | Die casting | Tooling lead time | Design freeze before tool cut |
| 10,000+ plastic | Injection molding | Mold cost and lead time | Wall thickness, draft angle |
Which route fits your batch
If the design is still moving, the material is hard, or the volume sits under a few thousand parts, run it as a CNC machining batch customized order with a dedicated fixture and scheduled tool changes. If the geometry is frozen and the volume is past 10,000 simple parts, tool up for casting or molding instead; milling it will cost more per part and take longer.
Questions we get on batch runs
How many parts before a dedicated fixture is worth building?
It depends on geometry more than count. For a simple prismatic part held in soft jaws, a dedicated fixture rarely pays back below 200 parts. For a thin-walled part where a vise bows the wall, the fixture pays back much earlier, sometimes at 30 to 50 parts, because the alternative is scrap.
We usually quote the batch both ways and show the crossover point. If the fixture adds cost now but removes a scrap rate, the arithmetic is easy to see.
Can you hold ±0.005 mm on a 500-piece run?
Yes, but not by leaving the machine alone. It requires temperature control in the shop, measurement with micrometers or a CMM rather than calipers, and a tool life plan with scheduled offset changes.
We publish ±0.005 mm as our working tolerance and run at a 99.99% qualification rate. If a feature needs to be tighter than that, we say so before quoting rather than after.
What is your minimum order quantity for a customized batch?
There is no minimum order quantity. We run from one prototype to 10,000+ part runs on the same floor, which is useful when you want a bridge batch before committing to tooling.
The practical floor is economics, not policy. A one-off part pays for setup; a 500-piece run spreads it.
Does tool wear really change the part size that much?
On a 6 mm carbide end mill in aluminium, flank wear of 0.02 mm is normal after a few hundred parts. That wear is transferred to the part, so the part gets smaller as the run goes on.
In 316L or Ti-6Al-4V, wear is faster, and the risk is not only size. A dull tool rubs, raises cutting temperature, and can tear the surface finish.
Do you provide inspection reports with batch orders?
Yes, on request. Our flow is a raw material check, in-process monitoring at defined part counts, and a final inspection before shipment. We inspect 100% of parts before they ship.
If your drawing calls out specific features, tell us at quote stage so the inspection plan matches the drawing rather than a generic check sheet.
How do you protect our design during a batch order?
Uploads are secure and confidential, and we sign an NDA on request before any file is reviewed. Files stay inside the project team and are not shared with other customers.
For long-running batches, we keep the fixture and program tied to your part number so a repeat order years later starts from the same setup.
Send the drawing, get a batch plan
Upload your files and we will return a quotation with a free DFM analysis within 12 hours, including the fixture and inspection approach for your volume.
12-hour quoteNo MOQ100% inspectionNDA on request