Pilot Build to Volume Production: Closing the Gap

Many programmes go straight from a working prototype to a production order, and the problems that were hidden by the small quantity then appear in the factory. A prototype is built in ones and twos, often by hand, with the attention of the designer on every board. A production lot is built in hundreds by a line that has to be able to repeat the same result without that attention, and the difference between the two situations is what a pilot build is for.

What the Pilot Build Actually Tests

The first purpose is to verify that the design can be manufactured. A layout that works in a prototype can still be difficult to panel, to stencil or to assemble, and a DFM review at the pilot stage finds the problem while the tooling is still being prepared. The second purpose is to test the stability of the product itself, using a quantity large enough for an ageing test, an environmental test and a reliability test to say something about consistency rather than about one board.

The third purpose is cost. Substituted parts can be compared, suppliers can be compared on both quality and price, and the material that will be used in volume can be chosen on evidence from the pilot rather than on a quotation alone. The fourth is the supply chain: the lead time and the stability of the critical components are measured before the volume order depends on them. The fifth is the process itself. The SMT parameters, the assembly sequence, the test method and the standard work instructions are established and proven during the pilot, which is what makes the later ramp a matter of scaling rather than of discovering.

A pilot of fifty to five hundred sets is the usual range, adjusted to the product. A consumer product may be sufficiently characterised at the lower end, while an industrial or medical product with more test content often needs the upper end, because the number of tests per unit is what determines how much evidence the sample produces.

Freezing the Design and Reviewing It

A pilot starts with a design freeze. Every change that has been agreed is in the released data, and the data that the factory receives is the data being built. When a change arrives in the middle of a build, the pilot stops being a single experiment and becomes two, and the results from the two halves cannot be combined.

With the design frozen, the engineering review covers the board and the bill of materials together: the manufacturability of the panel and the stencil, the testability of the nets that need to be reached, and the parts that could be replaced by an equivalent with a better price or a shorter lead time. That review is where a problem costs a conversation instead of a scrapped lot.

Optimising the Bill of Materials and Buying the Material

BOM optimization follows the review. Alternative parts are validated against the requirement rather than against the package, the parts that exist in more than one package are considered for standardisation, and the price and the availability of each critical part are compared across sources. Parts with long lead times are identified early so that a decision is taken deliberately rather than under pressure.

Once the bill of materials is fixed, the material is bought against it. Having a part qualified during the pilot and then buying from a different source in volume is a common way to lose the benefit of the pilot, which is why the qualified source is recorded rather than remembered.

Fabrication, Stencils and Assembly

Fabrication and stencil making run in parallel. Boards from one to thirty layers are produced in house, including high layer count, high density interconnect and high frequency material, and the stencil apertures are designed around the component packages so that the printing stage begins from a considered opening rather than from a default.

Assembly follows with the pilot treated as a production run rather than as a demonstration. Experienced engineers follow the first article and the process parameters, and the through hole parts, the conformal coating and the packing are handled in the same sequence that volume production will use. A pilot that is built with methods the volume line does not use proves nothing about the volume line.

Testing, Ageing and the Report

After assembly the boards go through in circuit and functional testing, which removes the joints and functions that are outside specification, and where the product needs it the units go on to a high temperature or a powered aging test that exposes the early failures before the customer does. The test data are recorded in enough detail to analyse the first pass yield and to identify the failure modes rather than only their number.

The pilot closes with a report that includes the yield analysis, the causes of the failures that were found, the changes recommended for the design, the process and the cost, and the risks identified in the supply chain. That report is what turns a pilot into a plan for volume production.

Carrying the Result Into Volume

The documents produced during the pilot travel with the product: the process parameters, the work instructions and the test specification. The suppliers and part numbers that were validated become an approved list rather than a memory. The yield measured during the pilot becomes the baseline against which production is judged, so that a drift is visible before the customer reports it.

A ramp plan is then built from the pilot data and the market forecast, so that capacity is arranged rather than discovered, and further design iterations are supported after the pilot if the product needs them. Our low volume assembly group runs the pilot, component procurement qualifies the material, and the same process continues into high volume assembly when the product is released.

Where a Pilot Is Not the Right Answer, and Where It Is Essential

A pilot adds cost and time, so it is worth being honest about when it does not pay. A product that is a variant of an existing one, built on the same panel, through the same process and with the same material, does not need a separate pilot, because the production line has already demonstrated that it can build the design. A board with a very low annual quantity may also be better served by building a slightly larger prototype batch than by running a formal pilot and then a separate volume order.

The cases where the pilot is essential are the ones where a failure would be expensive and late. A new package type that the line has not placed before, a stack-up with a controlled impedance that has not been produced at this fabricator, a change of surface finish, or a product that will be built in tens of thousands per month all belong in that group. So does any programme where the product has to be certified, because the certification requires evidence that only a batch of known size and known process can provide.

The middle ground is a pilot that is deliberately smaller than the eventual production quantity but built with the production process from the first board. It carries most of the information at a fraction of the material cost, and it leaves the larger validation to the first production lot, which is then treated as a controlled build rather than as an uncontrolled one.

Whatever size is chosen, the value comes from the build being representative. A pilot that uses a different stencil, a different inspection plan or a different profile produces data that describe the pilot rather than the product, and the factory then has to discover the production behaviour a second time.

FAQ

How large should a pilot build be? Fifty to two hundred sets suits a consumer product, and two hundred to five hundred suits an industrial or medical product where more test evidence is needed per unit.

How long does it take? Two to four weeks from the design freeze to delivery is typical, shorter where the material is in stock and the board is straightforward.

Can the volume order follow at the same factory? Yes, which removes the cost of qualifying a second supplier and of transferring the process.