Scaling a Verified Design into Volume Manufacturing
There is a point in every hardware programme where the design stops being a project and becomes a product, and that point is not the design freeze. It is the moment the design is released into volume manufacturing, where the boards are built by a line rather than by people who know the design, and where the questions are about consistency rather than about function. The work that sits between the two is the subject here, and it is mostly administrative: the settings, the records and the agreements that let somebody else build the same board the same way.
What the Pilot Run Is For
A pilot run is often treated as a smaller version of the production order, and that description misses its purpose. The pilot exists to exercise the process on the product, which means using the production stencil, the production programme, the production profile and the production fixture, on a quantity large enough to reveal a systematic problem and small enough to be investigated. A pilot that is built by hand or with the development equipment proves nothing about the line.
The questions a pilot is expected to answer are specific. Does the paste print consistently across the panel? Does the placement programme produce a uniform result or does one region show an offset? Does the reflow profile produce joints that meet the visual criteria on every position? Does the fixture make reliable contact with every board, and does the test programme distinguish a good unit from a marginal one? Does the packing survive the handling it will receive?
Each of those questions has an answer that is measured rather than observed. A pilot that ends with a statement that the boards look good has confirmed very little, and a pilot that ends with a set of numbers has established the baseline against which the production batches will be compared.
First Article and the Release Decision
The first article confirmation is the formal moment at which the setup is approved. The first board or the first panel is examined against the assembly drawing and the bill of materials, the paste condition is reviewed against the print specification, the placement and the orientation are verified, the joints are inspected and the separation method is tested. Where the order includes programming and testing, the first article continues into those steps rather than stopping at the appearance.
The release decision that follows is a decision about the process rather than about the board. A first article that is acceptable confirms that the stencil, the programme, the profile, the material and the fixture are all correct for the revision being built. That is why the record of the first article is retained: it establishes what the process was when the product was known to be good.
Where something is not acceptable, the correction is made and the first article is repeated rather than the batch being released with a note. A pilot run is the last point at which a systematic fault can be corrected without consuming material.
Process Capability and Consistency
Consistency is the characteristic that separates a pilot from a production order. A process capability describes what a process produces across many repetitions, not what it produced on the day it was examined, and establishing it requires measuring the same characteristic on a number of boards rather than on one.
The paste volume across a panel, the placement offset of a fine pitch device, the reflow temperature at a fixed position and the measured impedance of a controlled net are all characteristics that can be tracked this way. Where the spread is comfortably inside the tolerance, the process is under control and the production run can proceed with routine checks. Where the spread approaches the limit, the process needs attention before the volume begins, because a process that is only just inside its window will cross the boundary as soon as anything in the environment changes.
The distinction matters commercially as well as technically. A product whose process capability is barely adequate will produce an acceptable pilot and an unpredictable production batch, and the cost of that unpredictability appears as rework, as scrap and as delays, all of which are avoidable before the first production order.
Yield and What It Tells You
The yield of a pilot run is the first honest number a programme receives, and it is worth examining before it is quoted. A yield that is high but produced by extensive rework is not a high yield; it is a process that is consuming labour to conceal a defect. A yield that is low and concentrated at a single operation identifies the operation to investigate. A yield that is low and scattered usually points to a material or an environmental cause rather than to a machine.
For a customer, the useful question is not the number alone but the composition of it. What proportion of the boards passed without intervention, what proportion passed after rework and what proportion was scrapped? The first of those is the number that will apply to the production order if the process does not change, and it is the only one that describes the process rather than the effort applied to it.
The Documentation That Makes It Stick
The handover from development to production is completed by the release of the documents that define the product. The artwork, the stack-up, the bill of materials, the placement coordinates, the assembly drawing, the stencil identification, the reflow profile, the programme revision, the test requirements and the packing specification all need a revision number and a home. The temporary notes created during development are retired at the same time, because an instruction that is still circulating after the release is an instruction that will eventually be followed.
The final element is the agreement about change. A product in volume will be modified, and the procedure for doing so — who proposes, who approves, from which batch the change applies, and how the superseded material is withdrawn — is what prevents two revisions of a product from existing on the line at the same time. Our high volume PCB assembly route takes these programmes, the prototypes that precede them run through rapid PCBA prototyping, the small quantities are built under low volume PCB assembly, the records are held with quality management, and the testing follows under PCBA testing.
Comparing the Batch with the Baseline
The numbers produced by the pilot run exist to be compared with the numbers produced by the production batches, and that comparison is what turns routine inspection into process control. The paste volume, the placement offset, the first article result, the test pass rate and the yield of each batch are tracked against the baseline, and a movement away from it is investigated while it is still a trend rather than after it has become a defect.
Without the baseline, each batch is judged on its own and a gradual drift is invisible, because a slow change never looks wrong on a single measurement. With the baseline, the same drift appears as a departure from a recorded value, and the question becomes what changed rather than whether anything changed. That is the whole difference between a quality system that reacts and one that anticipates, and it costs nothing beyond keeping the records that the pilot already produced.
FAQ
How large should a pilot run be? Large enough that a systematic defect appears in more than one unit, and small enough that every board can be examined rather than sampled.
What is the difference between a pilot and a first article? The first article confirms the setup on one board; the pilot confirms that the setup produces consistent results across many.
When is a product ready for volume? When the process capability is comfortably inside the tolerance, the yields are understood and the documentation is released with revision control.
