CNC Verification: A Quick and Accurate Prototyping Guide
CNC verification is what happens between a released CAD model and the first cut chip. This guide explains the checks that matter, the parameters behind them, and when verification is worth the extra hours. Written for design engineers and sourcing teams ordering prototypes.
What CNC verification actually checks
CNC verification is the set of computer and machine checks that confirm a toolpath will cut the intended geometry before metal is removed. It is not a single command in CAM software. It is a chain: tool assembly data, stock definition, simulation, fixture clearance, and on-machine probing.
The reason it exists is simple. A 3-axis roughing pass can remove 40 mm of material in seconds. If the toolpath is wrong, the part is scrap and the schedule slips. Simulation costs minutes; a re-machined billet costs days.
Verification answers three questions. Does the tool reach every surface? Does it collide with the fixture or the table? Will the finished part satisfy the drawing tolerances, including the ones the probe must confirm?
A common misunderstanding is treating verification as a formality that only matters for complex 5-axis work. In practice, a simple 2.5D bracket with a deep pocket and a 3 mm corner radius fails just as often, because the chosen tool cannot reach the corner and the CAM operator does not notice until the part is on the bench.
How verification works, layer by layer
The first layer is the CAM simulation. The software loads the actual holder, collet, and cutter geometry, not a generic cylinder. It then steps through the program, subtracting swept volumes from a virtual stock block. Material removal is checked against the target solid.
Gouges show up as red regions where the tool cut too deep. Undercuts show up as grey regions the tool never reached. Both are reported per operation, so the programmer knows whether to change the tool, the stepover, or the workholding.
The second layer is machine simulation, which adds the kinematic model: table travels, rotary axes, spindle head, and fixture. This is where interference appears. A long tool holder may clear the part but hit the trunnion on a 5-axis machine at a 45° tilt.
The third layer is on-machine verification. After setup, a touch probe or laser tool setter measures the datum and checks critical features. On a prototype with a ±0.005 mm tolerance, probing before finishing cuts is cheaper than discovering a 0.03 mm shift after the final pass.
What verification cannot catch
Simulation cannot predict thermal growth. A spindle running at 12,000 rpm for 40 minutes heats the casting, and the Z axis drifts. On a tight-tolerance part, that drift can reach 0.01–0.02 mm. The fix is process control, not software: warm up the machine, keep coolant steady, and measure between operations.
It also cannot predict material behavior. A 7075 aluminium pocket walls move less than a 316L stainless wall of the same thickness, because the specific cutting force differs. Simulation uses a rigid model. Real parts deflect.
Fixture compliance is another gap. A part held on three points will ring during a heavy pass even if the toolpath is perfect. Add support or reduce depth of cut.
None of this makes verification unnecessary. It means verification is the first filter, not the last. Engineers still need a measurement plan for the features that matter.
Cost and time in prototyping
The time cost of verification is small. CAM simulation of a typical prototype runs in 5–20 minutes. Machine simulation with a full kinematic model can take 30–60 minutes, which is why it is reserved for 5-axis and complex setups.
The saving is larger. A scrapped prototype means a new billet, a new setup, and a new schedule slot. On a 3–5 day prototype turnaround, one scrapped part can push delivery past the deadline.
For quoting, verification effort is not a line item. It is part of process planning. What matters to the buyer is that the shop runs it as standard practice, not as an upsell.
There is also a documentation benefit. Simulation reports and probe records give the engineer a traceable record of what was checked. For medical and aerospace prototypes, that record supports the design review, even when the part itself is not the final production item.
Step by step: running verification on a prototype
Typical sequence for a first-article prototype
- 11. Fix the CAD and the datumFreeze the model revision. Define the datum from the drawing, not from the model origin. Every later check references this datum, so a change here invalidates all downstream measurements.
- 22. Build the tool assembly accuratelyLoad real holder and cutter dimensions. A 6 mm end mill with 20 mm of flute length behaves differently from a 6 mm cutter with 50 mm of reach. Reach is the number that causes collisions.
- 33. Simulate with a stock modelStart from a stock block that matches the actual billet, including saw-cut faces. Simulating from the finished part hides the first operation and misses unexpected material.
- 44. Check fixture and table clearanceVerify clamp positions, vise jaws, and rotary table limits. On a Ø400 mm rotary table, a part that is 380 mm long leaves little room for the tailstock.
- 55. Review thin-wall and chatter riskFlag walls thinner than 1 mm. Reduce radial engagement to 5–10% of cutter diameter and raise spindle speed to keep chip load stable.
- 66. Probe the datum on the machineTouch off X, Y, and Z, then confirm one known feature. If the probe result differs from the CAM assumption by more than 0.02 mm, correct the work offset before cutting.
- 77. Cut a first-article checkMachine the critical features, measure them, then release the rest of the program. For prototypes with tight tolerances, this catches setup errors before the full cycle.
Verification depth by part type
Match the check level to the risk, not to the machine count
| Part situation | Minimum verification | Why |
|---|---|---|
| 2.5D plate, open pockets, ±0.1 mm | CAM simulation, tool list check | Simple geometry, low collision risk |
| Deep cavity, L/D ratio above 4 | Simulation plus holder collision check | Long reach tools are the main failure mode |
| 5-axis contoured surface | Full machine simulation with rotary axes | Tilt moves can hit trunnion or table |
| Thin wall under 1 mm | Simulation plus cutting parameter review | Deflection, not collision, drives scrap |
| Tight tolerance ±0.005 mm | Simulation, probing, first-article check | Setup shift is larger than the tolerance |
| One-off prototype, 3–5 day ship | Simulation plus probe datum check | No time for a second setup trial |
When to verify more, and when to skip
If the part has 5-axis contours, thin walls, or a tolerance tighter than ±0.02 mm, run full machine simulation and probe the datum before cutting. If it is a simple 2.5D prototype in aluminium with ±0.1 mm tolerance, CAM simulation plus a tool list check is enough. Spend the hours where the risk is.
Common questions about CNC verification
Does CNC verification slow down a prototype order?
It adds minutes, not days. CAM simulation for a typical prototype takes 5–20 minutes. Machine simulation for 5-axis work can take up to an hour, but it usually prevents a re-cut that would cost 1–2 days.
At GreatLight, quotation and DFM analysis are returned within 12 hours, and production can start within 24 hours. Verification runs inside that window.
What materials can be verified this way?
Simulation is material-agnostic because it models geometry, not physics. It applies to aluminium grades such as 6061 and 7075, stainless 303 and 17-4PH, tool steels, titanium TC4, and plastics like POM and PEEK.
Material does affect the cutting parameters you load into the simulation. Harder materials need lower chip load and more conservative stepover, and those values change the result.
Is verification the same as inspection?
No. Verification checks the plan before cutting. Inspection measures the part after cutting. They are separate stages and both are needed on tight-tolerance work.
GreatLight inspects 100% of parts before shipment, with raw material checks, in-process monitoring, and final inspection. Reports are available on request.
How do you verify a part that is too large for the machine simulation?
Break the program into setups and simulate each one separately, then verify the transitions. On our 4,000 mm maximum processing size, long parts are often machined in two operations with a repositioned datum.
The risk moves to the re-datum step. That is where probing matters most.
What tolerance can verification realistically support?
Our machining tolerance is ±0.005 mm (±0.0002 in) on features that the process can hold. Verification helps reach that number by catching setup and toolpath errors first.
It does not remove thermal drift or fixture deflection. Those are controlled by warm-up, coolant strategy, and support placement.
Can I get the simulation and inspection records?
Yes. Inspection reports are available on request. NDA is available on request, and uploads are kept secure and confidential.
For regulated programs, tell us at quote stage which records you need so they are built into the process plan, not added afterward.
Send your prototype file for a verified quote
Upload your CAD and drawing. We return a quotation and free DFM analysis within 12 hours, and production can start within 24 hours. No minimum order quantity, from one prototype to 10,000+ parts.
12-hour quoteDFM feedback included100% inspection before shipmentNDA on request
More machining notes
We publish setup notes, tooling trials and inspection data from the factory floor.
