Presentation of the Prototype: What Engineers Should Actually Inspect

• 1
  What the Presentation of the Prototype Really Means
• 2
  Datum Strategy Decides Whether the Numbers Mean Anything
• 3
  Material Choice Changes What the Prototype Can Prove
• 4
  Fit, Stack-Up and the Things a CMM Cannot See
• 5
  Can the Process That Made One Repeat Ten Thousand
• 6
  Which Prototype Method Fits Which Question

A prototype is one or more functional samples cut from the same material and by the same process family as the intended production part, made without a mold. The presentation is the review where that sample is compared against the 3D model, the 2D drawing and the assembly it has to fit into. It is a technical event, not a marketing one.

The purpose splits into three jobs. First, check geometry: does the physical part match nominal within the stated tolerance, and where does it not. Second, check function: does it assemble, seal, rotate, snap or conduct as intended. Third, check process: can the shop that made the sample repeat it at volume without changing the setup logic.

Most prototype reviews fail for a boring reason. The team brought a part but no measurement plan. Someone holds a caliper against a bearing bore, gets 0.03 mm of scatter, and the room starts debating the design instead of the process. Bring the numbers you already promised in the drawing.

A prototype can measure perfectly and still be wrong. The cause is usually the datum reference frame. If the drawing calls A-B-C and the inspection report used the machined face as a single datum, the two sets of numbers describe different parts. Fix the datum scheme before you cut metal, not after.

For a first article from a 5-axis setup, we normally establish the primary datum on the largest stable face, the secondary on a pair of dowel holes, and the tertiary on a slot. That gives the CMM a repeatable origin. Position callouts then have a chance of landing inside ±0.005 mm instead of scattering across 0.05 mm.

Watch the difference between a model-based definition and a 2D drawing that was never updated. When the two disagree, the drawing usually wins in a review and the model wins in the machine. Pick one authority, write it on the traveler, and stop the argument at the source.

Thin walls and long cantilevers deserve their own note. A 1.5 mm aluminium wall will move after clamping release, so a measurement taken in the vise is not the measurement the customer needs. Measure free-state, then measure again on the assembly fixture.

The material you prototype in sets a hard ceiling on what the sample can demonstrate. Aluminium 6061-T6 is the default for housings and brackets because it machines fast and holds ±0.005 mm without drama. It will not tell you anything useful about wear, fatigue or galvanic behaviour.

When the production part is stainless, prototype in the same grade. Cutting 17-4PH in the H1025 condition at Ra 0.8–1.6 μm is a different job from cutting 6061, and the tool path, chipload and finishing pass all change. A prototype made in aluminium and approved for a 316L part hides real risk.

Titanium TC4 (Ti-6Al-4V) and Inconel are worth prototyping in the real grade when wall thickness is under 2 mm. Both work-harden and both spring back, so the machined geometry that comes off the machine is the geometry you need to see. Substituting steel tells you almost nothing about deflection.

Plastics behave differently again. POM and PEEK hold tolerance well but absorb moisture; ABS and PC print or machine with different shrink. If the final part is injection molded, a machined POM prototype proves fit, not moldability.

Functional checks beat dimensional checks when the part has to move. Bolt the prototype into the mating assembly and cycle it. A hinge that passes a CMM report but binds at 40 °C is a failed prototype, and the presentation is where that gets found.

Stack-up is the usual culprit. Each part sits inside tolerance, yet the assembly does not close because the tolerances accumulate in one direction. Run a worst-case and a statistical stack on the drawing before the review, then confirm it against the physical assembly.

Sealing faces and O-ring grooves deserve a pressure or leak check, not a visual. A groove depth that is 0.02 mm shallow will pass a caliper and fail at 2 bar. Bring the test rig to the review if the part has a sealing function.

For moving parts, record the torque or force needed to actuate the prototype at the start and after 500 cycles. That single number often decides whether the design moves to tooling or goes back for a wall thickness change.

A prototype cut on a 5-axis center with a custom fixture proves the geometry, not the production route. Ask early which process will make the real part: casting, molding, stamping or machining. Each has its own draft, radii and wall rules.

Machined prototypes rarely translate directly to die casting. A casting needs 1–2° of draft and uniform walls, so a machined sample with vertical faces and a 6 mm rib will not cast. Catch that at the presentation stage and the tool design changes once instead of three times.

If the production route stays CNC, the review should confirm the setup count. A part that needs four setups on a 3-axis machine may drop to two on a mill-turn center with a Ø400 mm rotary table. Fewer setups means fewer datum shifts and a tighter position result.

Surface finish belongs here too. A hand-polished prototype at Ra 0.2–0.8 μm can mislead a team into expecting that finish everywhere. Specify the finish per face, and state which faces are as-machined at Ra 1.6–3.2 μm.