A prototype run is the first time a design meets real material, a real line and a real operator. The drawings that passed internal review suddenly have to answer questions nobody asked in the meeting room: which film, which holder, which recipe, which tolerance can the workshop actually repeat. Writing those answers down before the order is placed is what separates a useful prototype from an expensive mock-up.
The list below covers the specifications we ask buyers to define when they request a trial production, and it applies whether the subject is a thin-film solar cell stack, an OLED or LCD module, or a small cleaning equipment unit being developed for a new market.
1. The workpiece definition, not just the drawing
Send the outline dimensions, the thickness of every layer, the material of each layer and which face is functional. A "5.5-inch display" is not a specification: two parts with the same diagonal can differ in curvature, total thickness and bending radius, and those three values decide whether a bonding process can run at all.
For bonded assemblies, list the stack from top to bottom, name each material, and state which surfaces must stay free of adhesive or bubbles. Where the sample is a customer-supplied material rather than one you can specify freely, say so — the machine settings follow from the incoming material, not from the target product.
2. The process window each step must sit inside
For every step in the trial, write down the parameter and its allowable range: temperature, pressure, vacuum level, alignment accuracy, cycle time. Write ranges, not single numbers. A specification that reads "2 t press pressure" without an acceptable band gives the engineer nothing to tune within when the first trial shows a defect.
Where a step needs a controlled atmosphere, state what has to be controlled. Nitrogen-filled chambers with oxygen monitoring, and vacuum levels held at a defined upper limit, are two different requirements with different costs; both appear in thin-film and debubbling work, and only one of them can be verified with a gauge if you do not say which.
3. How quality will be judged on a part, not on a report
Define the pass criteria in observable terms: permitted defect size, permitted number of defects, the area that counts as critical, and the inspection distance and lighting used to judge it. Alignment error, bubble size and fitting error need numbers, and those numbers need a measurement method. Otherwise supplier and buyer will disagree about the same part after the trial, which is the most expensive way to discover a missing specification.
Also state what happens after inspection. If trial parts must survive a downstream step — washing, lamination, a second bonding pass, field use — write that step into the specification so the trial proves something about the real process rather than about the machine alone.
4. Volume, cadence and what the prototype has to prove
Say how many pieces the trial should produce, over how long, and whether the point is feasibility, yield or speed. A ten-piece feasibility run and a run that must hold a stated pieces-per-minute rate drive different equipment choices, and the difference is not visible in an enquiry that only says "prototype". For a line where throughput is the objective, record the target cycle time together with the availability you expect from the equipment.
If the same machine must later cover a second product size or a curved surface, mention it in the specification. Changeover between sizes is a design decision, and a prototype built for one size only tends to lock that choice in.
5. Supply, utilities and the environment the machine will live in
List what is available at the installation site: supply voltage and frequency, compressed air, water source and connection, floor area, load limits and whether the machine is expected to move between positions. A bench trial that works in a laboratory still fails on a site where the power supply or water connection does not match, so these are part of the prototype specification from day one.
Name the consumables and the media you want to use, including the water quality and any cleaning agent. On a fixed steam cleaning machine, for instance, the water inlet type and the tank material belong in the specification just as much as the temperature and pressure, because both are constrained by what the site can supply.
6. Documentation, handover and the questions to ask back
Decide in advance what you want out of the trial besides parts: a parameter sheet with the values that worked, a report of the settings that failed, photographs of typical defects, or a written statement of which step limited the yield. Ask for these explicitly; a trial that produces parts but no recorded process window has to be repeated.
Expect the manufacturer to ask for what you may not be able to answer yet, such as the material batch to be used, the target line speed and the acceptance method your QC team actually runs. A development and testing service is most useful when those open points are surfaced during the prototype, not after it.
A short checklist to send with the enquiry
- Part: outline dimensions, thickness, layer stack, materials, functional faces
- Process: parameter with an allowable range for every step, plus atmosphere or vacuum requirements
- Quality: measurable pass criteria, defect size and count, inspection method
- Volume: number of pieces, time span, whether the proof is feasibility, yield or rate
- Site: power, utilities, floor area, consumables, water quality
- Handover: parameter sheet, failed settings, defect records, next-step recommendation
Put these six blocks in writing before the trial begins. The specification is what lets a prototype run answer a production question, and it is the document both sides can point to when the next step is decided.
