How to Take a Hardware Prototype to Production

A working prototype gives you evidence that an idea can function. Production preparation adds evidence that the intended product and build process can deliver acceptable results without the original developer fixing each unit by hand.

This guide is for founders and engineering teams who already have a functional prototype. It explains what to review, what evidence to collect and when to commit to the next build. The examples and calculations are illustrative; they are not supplier quotations or reported project test results.

After the prototype works, the next questions are repeatability, test access, and how the next units will be built.

The next stage is driven by repeatability. A development board, hand-cut enclosure or temporary cable may be useful during experimentation. Each becomes a production decision when you must specify how another person will build, inspect and service the product.

Stage names vary between organisations. Use them to identify the question being answered, rather than treating a label such as “production ready” as evidence by itself. A pilot that uses provisional tools or materials should state which production assumptions remain unverified.

Start by creating a baseline of what exists. Record the mechanical CAD revision, PCB revision, firmware version, BOM, assembly method and known failures. If one unit differs from another, document those differences before you interpret test results.

Then define the next build in practical terms. What is the intended quantity? Who will assemble it? What environment will it operate in? Which features are essential? What unit cost and delivery constraints matter? A batch for supervised customer trials has different evidence needs from a product sold for unattended use.

Turn broad requirements into checks. “Reliable connectivity” is difficult to approve. For a hypothetical monitoring device, “recover its connection within 30 seconds after the network returns, without losing stored readings” is testable. The recovery time must come from the product requirements, not from a convenient generic target.

A useful engineering review is a workaround log. Record every time someone trims a housing, bends a connector, holds a wire clear of a screw, chooses a special component or reloads firmware to make a unit work. For each workaround, choose a design change, a documented assembly operation or an explicit control. Unrecorded adjustments make later defects difficult to reproduce.

The output of this stage should be a gap list with an owner, an action and evidence required for closure. Agree the next phase before ordering tooling or a larger batch.

Mechanical and electronics decisions affect each other. A connector can move the enclosure opening. A stronger mounting boss can obstruct a component. A metal housing can change thermal behaviour or radio performance. Review the assembled product, including wiring, fasteners, antennas, seals and service access.

If a plastic enclosure will be injection moulded, review draft, wall thickness, ribs, bosses, undercuts, gates and ejection before committing to tooling. A shape that can be printed is not automatically suitable for moulding. Protolabs describes how uneven wall sections can contribute to sink and warpage, and why draft should be considered early when a printed or machined prototype will later be moulded. Final dimensions and draft depend on the resin, geometry and supplier process.

Apply the same reasoning to sheet metal and CNC parts. Check bend access, fastener access, tool reach and inspection points with the chosen supplier. Specify tight tolerances where a functional interface needs them; making every dimension tight can add cost without improving performance.

A board that can be debugged on a bench may be difficult to test once installed. Plan programming access, supply measurement, relevant test points, fixture clearance and diagnostic firmware before the layout is released. TI’s design-for-test analysis illustrates the link between physical probe access, board area and fault diagnosis; the test strategy must be adapted to the actual board.

Record exact component part numbers, approved alternatives and the conditions under which alternatives can be used. Matching a package or headline rating does not establish electrical, thermal or firmware compatibility. Any substitution that affects performance needs an engineering review and appropriate verification.

Ask someone who did not design the product to assemble it using the draft instructions. Watch where they hesitate, need an extra hand or cannot reach a fastener. Check connector orientation, cable routing, torque instructions and the order in which parts become inaccessible. Capture the cause, not only the elapsed time.

For example, a cable that is repeatedly pinched between the lid and base may need a routing feature or restraint. A warning in the work instruction can help, but it should not substitute for a practical design change when one is feasible.

Define acceptance criteria before the test begins. A useful test record identifies the unit, hardware and firmware revisions, equipment, conditions, method, measurements, pass limits and outcome. Photographs and logs make the result easier to investigate later.

The following matrix is a starting point for an indoor connected device. It is not a certification protocol. Select limits, duration and sample sizes from your requirements and risk assessment, and identify applicable market requirements with a qualified test partner early enough to influence the design.

Keep development testing separate from end-of-line testing. Development testing investigates whether the design meets requirements. End-of-line testing checks each manufactured unit for selected assembly and functional defects. A short production test cannot demonstrate the complete service life of every unit.

When a test fails, preserve the failed configuration, investigate the cause, update the affected design or process and repeat the relevant checks. A change to a housing, component or firmware build may also require regression checks on functions that previously passed.

The lowest quoted part price may not produce the lowest project cost. Compare tooling, engineering, fixtures, inspection, assembly, scrap, rework and packaging as well as parts. Confirm what each quote includes, its revision and the quantity assumptions.

Consider a hypothetical enclosure with two acceptable manufacturing routes. Assume route A has INR 15,000 in setup cost and INR 450 per enclosure. Route B has INR 2,50,000 in tooling and setup and INR 110 per enclosure. These are invented teaching assumptions, not current Indian market rates. Assume the same acceptable finished function; exclude tax, freight, design changes, testing, assembly and rejects.

Subtract the fixed costs: INR 2,50,000 minus INR 15,000 gives INR 2,35,000. Divide by the unit saving of INR 340. The simplified cost crossover is about 691 units, so route B becomes cheaper at 692 units under these assumptions.

That crossover is a decision input. If the enclosure will change after customer trials, a cheaper tool-based route may lock you into an immature design. Use confirmed volume, change risk, material performance, capacity and cash requirements alongside the calculation. Recalculate with comparable supplier quotes and expected yield before committing.

A pilot should exercise the intended assembly and test process. Use representative parts, materials, fixtures, work instructions and operators wherever possible. Record any departures from the planned production process so that the results have a clear boundary.

Track first pass yield separately from final acceptance. If 20 units enter the process and 16 pass all required checks without rework, first pass yield is 80 percent. Repairing the other four units and eventually accepting all 20 does not change that first pass result. Record defect categories, rework time and scrap so the team can act on the causes.

A small pilot also provides limited statistical confidence. If the true defect probability were 1 percent, the probability of observing zero defects in 20 independent, representative units would be approximately 81.8 percent. A clean batch of 20 therefore does not establish a defect rate below 1 percent. Repeated tests on the same unit are not 20 independent production samples. NIST provides exact binomial confidence methods for formal treatment of small samples.

Prioritise problems by their effect on safety, function, yield, assembly time and service. Repeated corrective work should lead to a controlled design or process revision. Build again when the change affects the evidence needed for release; record which tests can remain valid and which must be repeated.

A STEP file gives a supplier geometry. It does not fully describe materials, finish, functional tolerances, assembly order or test acceptance. Release a coordinated package whose documents refer to the same approved revisions.

• Mechanical package: native CAD where agreed, STEP files, dimensioned drawings, materials, finishes, critical tolerances and assembly details.
• Electronics package: schematics, board fabrication outputs, assembly data, BOM, programming procedure and firmware release information where included in scope.
• Build package: assembly instructions, cable routing, fastener requirements, fixtures, inspection points and approved test limits.
• Traceability package: unit identification, relevant hardware and firmware revisions, calibration records where required and test results.
• Change control: approved alternatives, revision history, deviation handling and responsibility for approving changes.

Ask the manufacturing team to review the package before release. Resolve gaps around incoming inspection, rejected material, repairs, packaging and service parts. Agree whether supplier changes need approval and who owns the released source files and tooling.

BrahmWorks has worked on the UltraFast charging hardware shown below. For a product of this type, manufacturing preparation involves reviewing the enclosure, screen mounting, cable routing, component access and assembly sequence as an integrated system. The applicable electrical safety and compliance work requires its own defined scope and evidence.

Use this checklist in a review meeting. For every item, record the evidence location, owner and unresolved action. A tick without an inspectable record is difficult to hand over.

• The intended use, operating conditions and measurable acceptance criteria are agreed.
• The next build quantity and intended manufacturing and assembly processes are defined.
• CAD, PCB, firmware and BOM revisions identify the same product configuration.
• Critical tolerances, materials, finishes and supplier capabilities have been reviewed.
• Programming, functional testing and fixture access are practical on the assembled product.
• Component alternatives are approved or have a defined verification path.
• Verification results are recorded and unresolved failures have a documented disposition.
• The pilot has build records, first pass yield, defect causes and rework information.
• Assembly, inspection, packaging and service instructions are available where required.
• Applicable certification, market and safety requirements have been identified and assigned.
• Manufacturing handover, source file ownership and supplier change approvals are agreed.
• Release responsibilities and the remaining actions are explicit.

Only when those deliverables are included in the agreed scope. Prototype development and manufacturing preparation may be separate phases. Define DFM, production drawings, supplier support, tooling, validation and pilot-build responsibilities before work begins.

The cost depends on the starting design, unresolved risks, intended manufacturing process, testing requirements and handover scope. Separate engineering fees from tooling, fixtures, parts, lab testing and pilot units. The enclosure comparison above illustrates one costing method; it is not a project estimate.

Build the schedule from the required design changes, supplier lead times, tooling, verification and pilot iterations. Add review dependencies and explain which activities can overlap. A universal timeline is not useful until the current prototype and release requirements have been reviewed.

Yes, if the agreed handover package, file access and ownership terms allow it. The receiving manufacturer should review the files against its own processes. A change in supplier, material or process may require updated verification or qualification.

If you have a working prototype, share the files you have, known failures, target quantities and operating conditions. BrahmWorks can help define the engineering work, validation and manufacturing documentation needed for the next stage. The review starts with your product and its evidence, then establishes phase-wise scope and deliverables.

Explore prototype to production support