A working prototype is easy to celebrate and easy to misread. The machine switched on, the operator pressed start, and something moved. That tells you the build is assembled. It does not tell you whether the unit can run a full shift, whether the protections trip when they should, or what your inspectors will check on every unit before it ships. Before you book test time, the practical question is not does the prototype work but which questions would change your next design decision.
Start with the production target, not the test list
Testing scope follows volume. A handful of units built by the engineering team and delivered with an engineer on site needs a different plan from the same design handed to an assembly line and a distributor. Put three numbers on the table first: how many units you intend to build, who will build them, and who will sign off each one. Those numbers decide how much evidence you need, because the cost of an undetected fault multiplies with every unit and with every party that touches the product.
If you have not decided on the production path yet, product R&D accelerator work usually runs alongside test planning, so the design and the test criteria are defined by the same team rather than argued about later.
Questions that size the prototype
Ask whether the prototype is built the way production units will be. Same enclosure, same fasteners, same cable routing, same supplier for the heating and pumping components. A prototype assembled by hand with bench wiring answers questions about function and almost nothing about assembly yield, serviceability, or vibration. Ask what is still missing from the prototype: no finish, no labels, no final control board, no production seals. Every gap is a question you will have to answer again later, at a higher price.
Then ask which modules exist as modules. In a steam and water machine, the heating path, the water path, the vacuum path, the electrical control and the mobile chassis can each be exercised on their own. Module-level views let you localize a fault instead of rebuilding the whole unit, and they let you retest one change instead of repeating an entire campaign.
Questions about function and performance
For each control on the panel, ask what should happen, what should happen if the operator gets it wrong, and what the machine should do if a sensor lies. Panel and HMI behaviour, actuators, alarms and protective functions are one family of questions. Performance is another: what pressure, temperature and flow does the unit actually deliver, measured rather than assumed, and how much energy does it take to deliver them? Cleaning effect is a measurable result, so write down how you will judge it before you test, otherwise every trial ends in an opinion.
Compare what you measured against what the datasheet for each part promises. Where they disagree, that is a design question, not a test error, and it is worth resolving before a purchase order for the next batch of components.
Questions about duty cycle and environment
Ask what a real day looks like. How many hours does the unit run between breaks, how often does it start and stop, what load does it carry, and what does the room look like: damp floor, ambient dust, cold garage, tight parking space, or a kitchen running all evening. Continuous running, repeated start-stop and high-load conditions expose parts that never fail in a ten-minute demo. Fault reproduction belongs here too; a fault you can trigger on purpose is a fault you can close out.
Ask how far the design is from a machine you already build. PowerDe's testing work concentrates on cleaning equipment series hardware — steam, water, vacuum and mobile platforms — plus automated and custom industrial products, so known-good platforms shorten the questions and put the effort on what is genuinely new.
Questions about safety and protection
With heated, pressurised, water-and-electricity equipment, the safety questions come before comfort questions. Electrical safety, leakage paths, emergency stop response, pressure and temperature limits, and the integrity of hoses, fittings and seals all need a defined pass or fail. Then ask what happens in the abnormal case rather than the normal one: empty water supply, blocked nozzle, blocked flow, a control signal that drops out. Protection functions are only real when you have watched them act.
Questions that prove longevity
Ask which parts wear out first and how you know. Seals, pumps, heaters, switches and buttons all have life expectations, and fatigue or wear shows up on hours, not on the first run. Life and reliability questions decide service intervals, spare-parts lists and warranty terms. Answer them from test records and you have a basis for those numbers; answer them from a datasheet alone and the number is a guess that a customer will discover for you.
Questions about production and factory testing
This is where most prototype reviews quietly fail. Ask what the assembly line will check on each unit, who records it, and what makes a unit a reject. A factory test plan needs the items, the method, the record form and the qualified or unqualified line written down before the first production batch. Ask the same questions about the manufacturing automation side: which steps are done by hand, which are automated, and what stops an out-of-spec unit from leaving the line.
Then ask what you hand to the next team. Test plan, test records, functional and performance reports, a problem list with causes, risk notes, retest confirmations, and optimization suggestions for structure, electrical control and factory inspection are the outputs that carry into trial production. A verbal debrief does not survive a production handover.
A short self-check before you commit
If you cannot state your production volume, the assembly owner and the sign-off rule; if you have no written pass or fail criteria; if the prototype is materially different from a production unit; or if nobody can reproduce the fault you are worried about — those gaps are your scope, and they are worth naming in the enquiry. Hardware testing work starts from exactly that information: current stage, product type, and the risk you care about most. Send those four facts and a test plan can be scoped against your product instead of against a generic checklist.
