PCBA Prototype: Building the First Boards Properly
A prototype build is the cheapest opportunity a programme gets. Design changes cost nothing while they are still in the CAD file, and cost progressively more at every stage after that: a stencil, a fabrication cycle, a material order, a delivered unit in a customer enclosure. Treating the prototype as a formality to get past is what turns a cheap decision into an expensive one.
What a prototype is actually for
A prototype answers questions that no simulation settles: whether the board assembles at the yield the design assumed, whether the thermal behaviour matches the model, whether the mechanical fit works in the real enclosure, and whether the firmware brings the hardware up. Those are the questions worth spending a build on. Verifying a price is not one of them, because the price at prototype quantity does not predict the price in production.
Prototype, pilot run and volume: three different jobs
| Stage | Typical quantity | Purpose | What it should produce |
|---|---|---|---|
| Prototype | 1 to 10 boards | Verify that the design works and assembles | A functional board and a list of design changes |
| Engineering build | 10 to 50 boards | Confirm the changes, run the product as a system | Frozen design, test coverage defined |
| Pilot run | 50 to 500 boards | Validate the process at production settings | Yield data, first article records, a production-ready process |
| Volume production | 500 upwards | Deliver product | Repeatable output and lot records |
The stages are not formalities, and skipping one is a decision rather than a shortcut. A programme that goes straight from prototype to volume without a pilot run has removed the opportunity to find a process issue on 50 boards instead of on 5,000.
Built on the production lines, not on a bench
Prototype and volume builds run on the same equipment and the same process here: the same four Yamaha high-speed SMT lines, three automatic DIP lines, the same 3D SPI before placement and AOI after reflow. That matters because a prototype assembled by hand or on a separate prototyping line can pass while the production process would not. When the prototype comes off the same lines as the production build, the result is representative, which is the only reason a prototype result is worth acting on.
What is separate is the setup rather than the equipment. A prototype carries a stencil of its own and a line setup of its own, which is where the one-time cost of a prototype comes from rather than from the machines.
What a prototype build includes
A prototype is a complete manufacturing run rather than a partial one. The same steps apply as on a production order, carried out at a smaller quantity.
- File review. Gerber files, BOM and assembly drawing checked for completeness, with anything missing listed rather than filled in with an assumption.
- DFM review. Panel utilisation, footprints, placement feasibility, component lifecycle risk and stack-up checked against the process before anything is ordered.
- Material plan. Components bought against the BOM through authorized channels, with long-lead and single-source parts identified before the build is scheduled.
- Stencil and line setup. A stencil cut for the design and a setup prepared on the placement line, which is where the one-time cost of a prototype sits.
- Assembly and inspection. SMT placement with 3D SPI before and AOI after reflow, through-hole and mixed-technology work where the design has it, and X-ray where joints are hidden.
- First article records. Inspection results and any deviation found during the build, issued with the boards rather than held internally.
What to settle at prototype stage
Most of the expensive corrections in a programme are decisions that were deferred past the prototype. These are the ones worth fixing while the build is still a prototype:
- Footprints and land patterns. A pad geometry that is marginal for the process will be marginal on every board of every future build. Fixing it at prototype costs an hour of design time.
- Acceptance class. Class 2 or Class 3 changes the inspection criteria and the records retained. Choosing it at prototype avoids building a prototype to one standard and a product to another.
- Surface finish and laminate. These set the storage window, the solderability and the fabrication route. Changing them later changes all three and invalidates earlier process work.
- Test strategy. What will be tested, and whether test points, probe access and fiducials exist on the board for it. Retrofitting test access to a released design is one of the most expensive changes available.
- Conformal coating and masking. Coating needs a masking plan derived from the assembly drawing. If the product is going to be coated, the prototype is when to define what has to stay clear.
- Programming route. Whether the board is programmed in-circuit, before assembly, or through a header, and whether a jig is needed. This determines access requirements on the layout.
- Board outline and panelisation. Department dimensions and break-off method affect both assembly handling and the bare board cost at volume.
Prototype lead time and cost
A prototype from stock material typically runs in 5 to 8 working days; where less common parts have to be bought, 2 to 3 weeks is more realistic. The MOQ and lead time page sets out the ranges by order type.
The cost structure of a prototype is different from production, and it is worth knowing why. The one-time items, a stencil, the line setup, engineering preparation and first article inspection, are incurred once and divided by a very small number of boards. That is the whole of the difference. Component cost does not fall because the quantity is small; it is simply not discounted. Reading a prototype unit price as a forecast of production cost is the most common misunderstanding in a first PCBA enquiry, and the cost page explains what changes as quantity rises.
Common prototype mistakes
- Leaving the BOM at description level. A prototype is where footprint and lifecycle problems surface, and neither can be found without manufacturer part numbers.
- Treating the prototype as a one-off rather than the first of a series. Design decisions taken for convenience on 5 boards can be carried into production without anyone deciding to.
- Not stating what the prototype is for. A board built to prove a circuit needs less test coverage than one built to prove a production process. Saying which it is changes the build plan.
- Assuming hand assembly is representative. It is not. A hand-built board tells you the design is electrically sound; it does not tell you the line can place it.
- Skipping the DFM review because the quantity is small. The DFM review is where marginal footprints, panel utilisation and obsolete parts are found, and its value is highest before tooling exists.
- Delaying the coating decision. Coating changes the masking plan and sometimes the layout. Deciding after the prototype means rebuilding it.
What we need to build a prototype
The same file set as a production build: Gerber files including drill data and board outline, a BOM with manufacturer part numbers and quantities, an assembly drawing showing orientation, polarity and keep-out areas, and the quantity. Where the design has already been built elsewhere and had yield problems, sending that history lets the review target the known issue. The quotation requirements page lists each item and what it changes.
Moving from prototype to production
The transition is mostly about converting decisions into fixed documents. The design version is frozen, the stencil and any test fixture already produced for the prototype are carried forward, the acceptance class and surface finish are fixed for the product rather than chosen per build, and the inspection records from the pilot run become the baseline the production build is measured against. Where the intended volume justifies it, the pilot run is the point at which the process itself is validated rather than the design.
Starting a prototype build
Send the file set to Gelant.lei@kilandu.com with a note on what the prototype is intended to prove. Prototype and volume builds run on the same lines and the same process, so a prototype result is representative of what production will produce. The turnkey PCBA page describes what a build includes end to end, and the ordering page covers the sequence and what each stage takes.
Can I order a single prototype board?
Yes. There is no fixed minimum order quantity, and a single board can be built. The unit price at that quantity is dominated by one-time items such as the stencil and the line setup, so it is not a useful forecast of production cost.
Is the prototype assembled by hand?
No. Prototype and volume builds run on the same SMT and DIP lines, with the same 3D SPI before placement and AOI after reflow. What is separate is the setup, not the equipment, which is why a prototype result is representative of what production will produce.
What should be decided at prototype stage?
Footprints and land patterns, the acceptance class, the surface finish and laminate, the test strategy and test point access, whether the board will be conformally coated and with what masking, the programming route, and the board outline and panelisation. These are cheap to change at prototype and expensive afterwards.
Do I need a pilot run before volume production?
It depends on the risk in the programme. Where the design is new, where the assembly carries fine-pitch or high-mass components, or where a previous build had yield problems, a pilot run validates the process on 50 boards instead of discovering the issue on 5,000. Where a design is already in stable volume production elsewhere, a first article from production tooling is usually enough.