CNC Machining Small Lot Parts: How Setup Decides the Result

• 1
  Why cnc machining small lot work behaves differently
• 2
  Setup and workholding for small lots
• 3
  Choosing toolpaths and parameters that suit low volume
• 4
  Inspection and tolerance on a short run
• 5
  Where small lot parts meet assembly
• 6
  What drives cost and lead time in a short run
• 7
  When small lot machining is the wrong choice
• 8
  Which approach fits the lot size and geometry

A small lot is not a scaled-down production run. It is a different economic and physical problem. At 10,000 pieces the cost of a fixture is spread thin, so you build a dedicated one. At 20 pieces the fixture has to be cheap and correct at the same time, and the operator has to find the datum on a part that may have no flat face.

The physics does not change: the tool still deflects, the material still moves when it is cut, and heat still walks the part around. What changes is how much of that variation you can average out. A production run averages it across thousands of parts and adjusts. A small lot has to predict it in the first article.

That is why cnc machining small lot quotes vary so much between shops. The machine hour is similar. The difference sits in setup planning, in whether the shop measures the first part properly, and in how quickly they can correct a drifting dimension without scrapping the whole batch.

For prototypes, bridge quantities and end-of-life spares, this is the whole game. The part works or it does not, and there is no second batch to fix it in.

• 1
  Setup dominates costAt low volume, fixturing and first-article time are often larger than cutting time.
• 2
  First part carries the riskA drifting dimension found at piece 40 means 39 suspect parts, not a fine-tuned process.
• 3
  Small features cut deeperA 2 mm end mill on a 3 mm wall deflects more than the same tool on a block.

The first decision is how many setups the part needs. Every extra setup adds a datum transfer, and every datum transfer adds a small error that stacks. For small lots we aim for one op-one-fixture wherever geometry allows, using 5-axis or mill-turn to reach features that would otherwise need a second vise position.

Soft jaws machined in place are the default for small lots. They are cut on the machine that will run the part, so the jaw profile matches the blank and the seat is true to the spindle. The cost is one short program and a few minutes of cutting. The gain is repeatability across 20 or 200 pieces without a dedicated plate.

When the part is thin, the vise is the wrong tool. A thin plate pinched in a vise bows, and the bow springs back after the clamps release. Vacuum chucks, potting in low-melt alloy or sacrificial tabs keep the part flat through the cut. The trade-off is handling time, which is acceptable at low volume.

For round or prismatic parts up to Ø400 mm, a rotary table lets one setup cover four faces. That removes three re-clamping steps, and with them three chances to lose the datum. It also keeps the operator from re-probing the part after every turn.

• 1
  One setup, one datumFewer re-clamps means less stack-up error across the batch.
• 2
  Soft jaws cut on the machineCheap, fast, and true to the spindle for the whole lot.
• 3
  Do not pinch thin wallsUse vacuum or potting when wall thickness is under about 3 mm.

Small lots reward a conservative first pass and a fast finish. On the first article we use a moderate stepdown, watch the chips, and listen for chatter before pushing. Once the part measures in, we can raise feed on the remaining pieces. That order matters: proving a toolpath on piece one costs minutes, finding a problem on piece thirty costs the lot.

Roughing with a larger tool and finishing with a smaller one is standard, but on small lots the finishing tool is often the constraint. A 3 mm end mill at 3× diameter reach is comfortable; at 6× it will deflect and the wall will taper. If the drawing calls for a deep narrow pocket, we either accept a tapered wall within tolerance or move to EDM or a smaller reach with a shallower depth per pass.

Aluminium 6061 and 7075 cut cleanly at high speed, so surface finish is easy to hold. Stainless 316 and 17-4PH work-harden, so a light rub instead of a cut will ruin the surface and the tool. Titanium TC4 (Ti-6Al-4V) needs lower surface speed and more coolant, and it moves when you remove material from one side.

Plastics are their own case. POM and PEEK hold dimension well if you keep the part cool; ABS and PP deflect under clamping and need sharp tooling and light chipload. Carbon fibre is abrasive and needs diamond or coated tooling plus dust control.

• 1
  Prove on piece oneValidate the toolpath before the batch starts, not during it.
• 2
  Watch reach-to-diameter ratioBeyond 6× diameter, expect taper and chatter on unsupported walls.
• 3
  Match material to methodWork-hardening alloys punish light rubbing cuts; plastics punish heavy clamping.

Achievable tolerance is a function of the feature, not of the shop. A flat face on a stable block in aluminium is straightforward at ±0.005 mm. The same callout on a 0.8 mm wall, 40 mm from the nearest support, is a different job. Before agreeing to a tolerance, we look at where the dimension sits relative to the datums and how much material surrounds it.

Temperature matters more on small lots because there is no long run to let the machine settle into a stable thermal state. A shop at 20 °C and a part measured at 20 °C agree. A part that is warm off the machine will measure small, then grow as it cools. We let critical parts normalize before final inspection.

Inspection is 100% before shipment, with raw material checks, in-process monitoring and a final inspection. Reports are available on request, including dimensional reports and material certificates. For a 5-piece lot that is the only control you have, so it should be thorough rather than sampled.

Surface finish follows the same logic. As-machined surfaces land around Ra 1.6–3.2 μm, a good finish sits at Ra 0.8–1.6 μm, and fine finishing reaches Ra 0.2–0.8 μm. Each step down costs time proportional to the area, which is why finish calls on large surfaces move the price more at low volume.

• 1
  Tolerance follows geometryThe same number is easy on a block and hard on a thin rib.
• 2
  Let parts normalizeMeasure at room temperature, not straight off the spindle.
• 3
  100% inspectionEvery part checked before shipment; reports on request.

Small lots often arrive as a kit: a housing, a bracket, a shaft and a cover that have to fit together. If each part is measured against its own drawing, the assembly can still fail, because two parts at opposite ends of their tolerance bands may not mate. The fix is to think in fits rather than in individual tolerances.

For a press fit, we hold the bore to the low side and the pin to the high side within the allowed band, and we check the pair with a gauge rather than with two separate reports. For a clearance fit, the opposite. This is a planning decision made at quote time, not something the machinist can rescue later.

When a small lot is destined for assembly, we ask for the mating parts or the fit class. If both sides are made in the same shop, we can assign the tolerance band across the pair and keep the assembly working. If they come from different suppliers, the drawing needs to state the fit, not just the nominal size.

Marking helps too. Laser marking with a minimum character height of 1.5 mm lets each part carry a lot number or a mating reference, so assembly errors are traceable instead of mysterious.

• 1
  Plan fits, not single dimensionsSplit the tolerance band across the mating pair.
• 2
  Gauge the pairCheck the fit itself, not two separate inspection sheets.
• 3
  Mark the partsLot or mating references at 1.5 mm minimum character height.

Per-piece price is high at low volume, and that is normal. The fixed part of the job, which is setup, fixturing, first-article inspection and programming, does not shrink with quantity. What drops is the fixed cost divided by more parts. Going from 5 to 50 pieces usually cuts the per-piece price sharply; going from 500 to 5,000 cuts it much less.

Material is the other lever. Bar stock and plate come in standard sizes, so a part designed around a standard extrusion is cheaper than one that needs a custom cut. On a 10-piece lot the material cost is small, but the wait for a non-standard size can be the longest part of the schedule.

Scheduling matters as much as machining. Production can start within 24 hours of an approved drawing, and parts typically ship in 3–5 days. That is possible because the shop runs 127 high-precision CNC machines, including 16 simultaneous 5-axis centers and 16 mill-turn centers, and because small lots can slot into gaps in the schedule.

The history here is a late-delivery probability below 2%. That number is not a promise for a specific order; it is what the shop has recorded over time. The practical reading is that small lots are treated as first-class work rather than as filler between big runs.

• 1
  Fixed cost dominatesSetup and programming do not scale down with quantity.
• 2
  Standard stock is fasterDesign around available bar and plate sizes.
• 3
  Small lots fit gapsThey can start within 24 hours and ship in 3–5 days.

If the part is a simple, thin-walled shell in the thousands, machining is the wrong process. Die casting, injection molding or vacuum casting will beat it on unit cost once tooling is amortized. Machining a small lot is about speed and geometry, not about the cheapest unit price at volume.

If the geometry needs a mold-like surface over a large area, or an internal channel that no cutter can reach, machining alone will not deliver it. That is a case for a hybrid build, where machining handles the critical faces and another process handles the rest.

If the tolerance callout is tighter than the geometry can support, no shop will hold it reliably, and a quote that claims otherwise is a quote that will end in a rejected lot. The honest move is to relax the callout, change the design to add support, or accept a different process.

Finally, if the material is exotic and the quantity is one, check availability first. Inconel and magnesium AZ31B are machinable, but the stock may take longer to arrive than the cutting takes to run.

• 1
  High volume, simple shapeCasting or molding wins on unit cost.
• 2
  Unreachable internal featuresMachine the critical faces, build the rest another way.
• 3
  Tolerance beyond geometryChange the design or the process, not the shop.