From Prototype to Production: What Changes on the PCB

The gap between a working prototype and a stable production board is wider than most teams expect. The prototype proves that the design functions. Production has to prove that the same design can be built thousands of times, by a different process, with materials that may not be the same, and tested in a way that catches the faults that the prototype build never had a chance to produce.

What Actually Changes

Five things change at the transition, and each one is a source of surprises.

The process. A prototype is often built on a small batch line with hand placement, a batch oven and generous inspection. Production uses stencils with different aperture design, a conveyor reflow profile, machine placement and automated inspection. The same design behaves differently under those conditions, particularly in paste volume, thermal profile and handling.

The panel. Prototypes are frequently ordered as single boards or in a simple array. Production panelisation is designed for the assembly line: it balances utilisation, adds rails and fiducials, and chooses a depanelisation method. A panel change invalidates the stencil, the assembly program and the fixture.

The materials. The laminate grade may be substituted for an equivalent, the surface finish may change to suit the volume process, and components may move to a different supplier or a different package variant. Each substitution is usually safe in isolation and occasionally not safe at all.

The test. A prototype is typically tested functionally, sometimes manually. Production needs a repeatable test with defined coverage, which usually means a flying probe or an in-circuit fixture, and that requires test points that the prototype may not have.

The documentation. Fabrication drawings, assembly drawings, BOM revision control and acceptance criteria all become formal. The prototype may have been built from an annotated email; production cannot be.

The Failures That Recur

Most production handover problems follow a small number of patterns.

The first is a tolerance that was never defined. A prototype works with a nominal value; production produces a distribution. If the design has no margin for the distribution, it will work in the lab and fail intermittently on the line. The fix is to identify the sensitive parameters early and to check them against the supplier capability rather than against the prototype units.

The second is a component change that was not reviewed. A part becomes unavailable, a substitute is found, and the substitute has a different thermal characteristic, a different footprint tolerance or a different leakage current. Every substitution should be recorded against the design and reviewed for its effect, not accepted because the pinout matches.

The third is test coverage that does not transfer. A prototype tested by an engineer with a scope cannot be tested the same way in production, and the faults that only appear in volume are the ones the production test does not cover. Designing the test points and the test strategy in parallel with the design, rather than after it, is the only reliable fix.

Design Freeze and Change Control

The transition works best when there is a clear point at which the design stops changing, and a defined route for the changes that turn out to be necessary.

A freeze is not a promise that nothing will change. It is a statement that a change now costs a revision, a document update and possibly a new stencil or fixture, so changes are made deliberately. That discipline is what prevents the common situation where three different builds of the same product carry three slightly different designs, and nobody can tell which one was qualified.

The mechanics are simple. One revision identifier applies to the board, the BOM and the assembly drawing together, and every document states it. A change is recorded with a reason, an impact assessment and an effective date. Units built before and after a change are traceable to the appropriate revision, which matters when a field issue appears two years later.

Two practical points are worth adding. First, the revision should be marked on the board itself where the space allows, because a board without an identifier cannot be traced to its documentation. Second, change control should cover the process as well as the design: a new stencil supplier, a different laminate source or a changed reflow profile all alter the product as much as a component substitution does.

Qualification and First Article

The first volume build is where the process is verified rather than the design.

A first article inspection checks the first boards off the line against the documentation: dimensions, hole sizes, copper thickness, finish thickness, mask registration, component placement, orientation and joint quality. Where the product has contractual requirements, the inspection usually follows a defined standard and produces a report that can be audited.

The second element is a capability check on the parameters that matter to the circuit. Impedance, for a controlled impedance board. Plating thickness in the barrels, for a board with a high aspect ratio. Solder paste volume and joint shape, for a fine pitch assembly. These are measurements of the process distribution rather than of a single unit, and they are what tell the customer whether the design will remain stable as the supplier runs more panels.

The third element is the test. The production test should be proven against known good and known bad units, including deliberately created faults if they can be made. A test that has never failed a board has not been shown to work.

Cost and Volume Curves

Cost behaviour at the transition is worth planning because it affects which decisions are worth making.

Prototype builds are priced by setup. Production builds are priced by material and machine time. The crossover usually happens somewhere in the low hundreds of boards, and the exact point depends on the layer count, the panel utilisation and the assembly complexity. Below it, paying a premium for speed is rational. Above it, the same premium becomes a permanent overhead.

That is why some decisions should be made twice. A part chosen for availability in the prototype may be the wrong choice in volume, and a surface finish that was convenient for the prototype may not be the one the assembly line prefers. Reviewing the BOM, the finish and the panelisation specifically for volume, rather than inheriting the prototype choices, is the step that most often removes cost.

It is also the point at which the supplier question should be asked again. A supplier who is excellent at prototypes may not be the right one for volume, and moving the product means transferring the design data, the process parameters and the test fixture. Doing that once, early, is far cheaper than discovering the mismatch after a production ramp has begun.

Transition Checklist

• Compare the prototype stackup and the production stackup and confirm that every difference is intentional.
• Confirm the production laminate grade and finish, and check the electrical and thermal effect of any substitution.
• Re-check component availability at volume, package by package, including the parts that look trivial.
• Fix the panelisation before the pilot, and keep it unchanged into production.
• Design test points for the production test method, even if the prototype does not use them.
• Freeze one revision identifier across board, BOM and assembly drawing, and mark it on the board.
• Run a first article inspection with dimensional and process measurements, not just a functional check.
• Verify the production test against known bad units before relying on it.
• Record the process parameters used, so a repeat build starts from a known state.
• Agree what will happen if the design has to change after the freeze, including who pays for the stencil and the fixture.

Test Strategy Across the Transition

The test method usually has to change, and the change is easier when it is planned.

A prototype is commonly tested with a flying probe or with manual bench checks, because there is no fixture and the quantities do not justify one. Volume production usually wants in-circuit test for structural coverage and a functional test for behaviour, because both are fast on every board.

Moving from one to the other means adding test points, which means a layout revision if they were not designed in. It also means creating a fixture, which is specific to the board and to the panel. Where the plan is known at the prototype stage, the test points can be included from the beginning at almost no cost, which is the single cheapest engineering decision in the whole transition.

Documentation

The documentation package is what allows a product to be built by someone who was not present during development. At minimum it contains the fabrication drawing with the stackup and tolerances, the assembly drawing with placement and polarity, the BOM with approved sources, the acceptance criteria for the finished product, and the process parameters that were used for the qualifying build.

Where the product is regulated, the same package carries the evidence that the requirements were met, and it is inspected rather than merely referenced. Even where it is not regulated, a package that a new supplier could build from is the difference between a transfer that takes two weeks and one that takes two quarters.

FAQ

• Can the prototype supplier build the production volume? Often yes, particularly for moderate quantities. What matters is whether the process capability matches the design, not the company size.
• Does the design have to change to go to production? Not the electrical design, but the panelisation, the test points and the documentation usually change.
• How long does the transition take? From a frozen design with a complete package, a few weeks. Most of the delay comes from component availability and fixture lead time.
• What is the biggest cause of production problems? Undocumented substitutions in material, finish or components, and a test that does not cover the faults the volume process can produce.

Summary

The move from prototype assembly to production is a change of process, not a change of design, and the failures that come with it are the failures of that process: undefined tolerances, unreviewed substitutions, a panel that invalidated the tooling, and a test that no longer sees the faults.

Handling it well is mostly administrative. Freeze a revision, keep one identifier across the documents, verify the first article against measurement rather than against opinion, incorporate the production test points in the design, and review the material and finish choices specifically for volume. Confirming the stackup and the capability with the fabricator, agreeing the test method with the assembly house, and checking the capability envelope before the design is frozen turns the transition into a scheduled activity instead of a series of surprises.