Prototype test plan template

1.Choose the prototype type

Different prototypes answer different questions, and each one costs differently. Pick before drawing or ordering anything.

1.1Six prototype types

1.2Sequencing principle

• One prototype, one or two questionsDo not stack. If a test fails, you won't know which assumption was wrong.
• Cheapest firstA foam-board appearance model can answer ergonomics that a $40 k functional prototype cannot answer any better.
• Don't skip stagesArchitecture before functional; functional before tooling; tooling before pilot. Skipping costs more than running.

2.Define the tests

A test is only useful if it has a method, a pass criterion, and a decision tied to the result. Write all three before building.

2.1Per-test definition (template)

2.2Standard environmental test methods

2.3Drop test heights by product class

2.4Calibration and controls

• Calibrate against the spec, not the previous prototypeEach test is judged against the brief.
• Document equipment with calibration certificateDate, traceability, accuracy. NIST/UKAS-traceable preferred.
• Record environmental conditions during testT, RH, supply voltage, anything that can move the result.
• Photograph the setup before startingSettles disputes 6 months later.

3.Run and review

Most prototype rounds fail because results aren't captured rigorously enough to act on. Treat the record as the deliverable.

3.1During the round

• Run tests in the planned orderDon't reorder by convenience; the order is usually riskiest-first.
• Record every readingEven ones that obviously pass. Trends across a batch tell you more than individual results.
• Photograph failure modesA blurry phone photo today is worth a long memory-based discussion later.
• Stop when a destructive test is reachedOr save those for last on units you no longer need intact.

3.2Test report template

`` Test ID: T-007 Question: Does the latch survive 500 open-close cycles? Standard: IEC 60068-2-79 (vibration), section adapted Method: Automated cycler, 30 cycles/min, ambient Sample size: n = 5 (units 003, 007, 011, 014, 019) Acceptance: Latch force after 500 cycles ≥ 80% of initial Result: Unit 003: 100% → 88% (pass) Unit 007: 100% → 92% (pass) Unit 011: 100% → 76% (fail; investigate) Unit 014: 100% → 84% (pass) Unit 019: 100% → 91% (pass) Mean: 86%; n=5; 4/5 pass; 1 outlier Disposition: Inspect unit 011 latch geometry; possibly material flaw or insertion variance. Re-run with n=10 if material change. ``

3.3MTBF / MTTF estimation

For reliability claims, you need multiple units × multiple hours.

`` MTBF (Mean Time Between Failures) = total operating hours / number of failures MTTF (Mean Time To Failure) — same formula; used for non-repairable items ``

Worked example: 10 units, run continuously, 3 failures over 90 days = 21 600 unit-hours / 3 failures = 7 200 hours MTBF (~10 months). To claim 50 000 h MTBF (~5.7 years), you'd need either: (a) 10 units × 5 000 hours each with no failures, or (b) much larger sample. Reliability claims need real data, not extrapolation.

3.4Review meeting

• Walk the test list end-to-endEvery line gets pass / fail / inconclusive. Inconclusive means re-test, not "ignore".
• Decide the design action before moving onFor each failed test, agree on the change and the owner. Open questions go on a parking list, dated.
• Update the briefIf the test result invalidates a brief assumption, update the brief now, not later.

3.5What to write down

4.Decide what comes next

The point of a prototype round is to enable a decision. Make the decision explicit before scheduling the next round.

4.1Before the next investment, confirm

• What changedIn CAD, PCB, BoM, firmware, assembly process, supplier choice.
• What needs to be re-testedAny change to a part touched by a previous test re-opens that test.
• What's ready to move forwardDecide explicitly which areas are stable enough for supplier quotation, certification samples, tooling, or pilot production.

4.2Gate criteria for common transitions

4.3Common round budgets