A prototype is the cheapest place to find out that a requirement was written wrong. On a perovskite programme, most of the expensive mistakes are made on paper: a coating window that was assumed rather than measured, a handling method nobody thought to specify, an interface between two stations that worked in the laboratory but not on a line. The purpose of a prototype is to convert those assumptions into numbers you can hold a supplier to. The steps below describe how to run that process so the prototype changes the specification instead of merely confirming that the machine switched on.
1. Write the requirement list before the prototype is built
A prototype without a written requirement list produces impressions, and impressions do not survive a purchasing meeting. For each requirement, write it as a sentence that can be tested: what must happen, on what material, in what sequence, and what measurement decides pass or fail. Requirements such as "good coating quality" cannot be tested; a requirement that names the layer, the substrate format, the target uniformity across the coated area and the method used to measure it can be.
Mark each requirement as one of three kinds: fixed by the process, fixed by the site, or still an assumption. The prototype exists to attack the third group first. If a supplier tells you a parameter is settled, ask which trial produced the figure.
2. Decide what the prototype has to prove
On perovskite production line equipment programmes the open questions usually fall into a short list: whether the coating method holds the film across the format you intend to run; whether the drying or annealing step reaches the temperature the process needs at the centre of the substrate rather than at the edge; whether the transfer between stations can be repeated without damage or contamination; and whether the sequence of stations can actually run together rather than one after another under supervision.
Name the two or three questions that carry the most cost if they turn out wrong. A prototype that answers those is worth more than one that demonstrates every station at a comfortable, unhurried pace.
3. Match the prototype's scale to the decision it supports
A bench-scale cell is a legitimate way to answer a materials question, and a poor way to answer a handling question. The failure modes that cost the most money on a line appear at full format: a substrate that flexes as it moves, a coating head that behaves differently at width, a drying step whose uniformity depends on airflow across a large area. Decide, requirement by requirement, what scale the test must be run at, and say so in the plan. Where the format is not yet fixed, use the prototype to fix it, and record that the format decision was made on prototype evidence.
For a programme that may move from coating to laser patterning and lamination, the prototype is also where you check whether a glove box spin coating equipment setup from development work can be carried into a line format at all, or whether the coating step has to be re-specified for the production method. That is a decision best made while the change is still a drawing.
4. Fix the trial design before the first run
Every trial needs a fixed starting point: the substrate and its cleaning state, the ink or precursor batch, the coating parameter set, the atmosphere in the enclosure, the drying profile, and the measurement positions. Change one factor at a time and record every factor, including the ones you think are irrelevant. A prototype that produces a good result nobody can reproduce is a demonstration, not a validation.
Write the parameter set on the sheet before the run, not afterwards. Runs whose parameters were written down after the result are the ones suppliers quietly improve when a customer asks for a repeat.
5. Measure with methods you agreed in advance
Agree the measurement before the trial: what instrument, at which positions on the substrate, how many samples per run, and what defines a reject. For coating uniformity that normally means a fixed set of positions across the format rather than a single centre reading. For a bonding or lamination step it means inspecting the same edges every time, because that is where entrapped particles and incomplete contact show up first.
Where the customer does not own the measurement equipment, the prototype is the point at which that gap is found. Ask the supplier to run the measurement method you intend to use in production, not their own preferred one, and keep the raw data rather than a summary chart.
6. Test the interfaces, not just the stations
Most line-level failures sit between stations. A coating station that works alone can still fail the line if it cannot hand a substrate to the next station in the state that station needs, in the time the line requires. Test the hand-off explicitly: the state of the part on arrival, the fixture or carrier it travels on, the time available, and who or what moves it. Where the process includes an inert atmosphere, check the transfer through the airlock as a trial in its own right, including what happens when an operator needs to intervene.
Run at least one full sequence end to end even if the individual stations were already proven. The end-to-end run is usually the first time anyone sees the whole process, and it is where the missing requirement shows up.
7. Verify the site conditions the machine will meet
A prototype is the right moment to check that the laboratory or supplier hall differs from your plant in ways that matter: power supply, compressed air, water, exhaust, gas supply, floor loading, door width, ambient temperature and humidity, and the cleanliness class of the room. Record what the prototype needed to run and compare it against what your building offers. If you do not have the services, that becomes a requirement in the specification rather than a surprise at delivery.
8. Ask for repeat runs and a clear failure record
One good run proves capability; three consecutive runs with the same parameter set prove control. Ask for consecutive runs under the same conditions and record every deviation, including jams, cleaning stops, and manual intervention. A supplier who cannot produce a failure record has either not been running long enough or has not been recording. Both answers matter to your programme.
9. Close the prototype with a changed specification
A prototype that ends without editing the specification has failed, however well it ran. Issue a revised requirement list with each line marked as confirmed, changed or unresolved, and attach the trial data that supports the change. Requirements that remain unresolved go on a short list with a named owner and a date, and they should be visible in the commercial discussion that follows, not buried.
The confirmed parameter set becomes the acceptance baseline for the production machine. Whoever ran the prototype should sign the sheet, and the same sheet should be handed to the team that will commission the line.
10. Decide what is developed and what is purchased
Prototype results often show that the coating or handling step is the risk, and that the remaining stations are standard. That finding lets you split the programme: develop the difficult step, and buy the proven ones. Where you need engineering support across the whole sequence, integrated engineering team services cover the process, controls and commissioning in one scope, and hardware testing can be booked as a defined stage with its own exit criteria. The wider equipment range is listed under thin-film solar equipment.
If you have a prototype plan and want it reviewed against what a line can actually deliver, send the requirement list and the trial data to the contact page and ask which of your open questions the prototype can answer and which it cannot.
