SMT Changeover and First Article Verification

A changeover replaces the stencil, the feeders, the programme, the support tooling and often the paste itself, and every one of those is a variable that can move the process. Treated as a single event, the changeover hides mistakes; treated as a list of independent changes with a verification attached to each, it becomes a controllable process. This article sets out that list and the measurements that close it out.

What a Changeover Actually Changes

A changeover changes the stencil, the feeder set, the machine programme, the support tooling and frequently the paste. Treating it as one event rather than a list of independent changes is what allows a single mistake to pass unnoticed until it appears as a defect.

The list is written per product, and each item carries a verification rather than a checkbox. A feeder that has been loaded but not verified is a risk that appears as a wrong or missing component many boards later.

The Setup Verification List

The items to confirm are the stencil identity and condition, the feeder positions with their part numbers, the programme revision, the support pin pattern, the paste type and lot, and the oven profile that applies to the product.

Each item is verified against the documentation for the product rather than against the previous job, because the previous job is often the source of the mistake. The verification is signed and recorded with the time and the operator.

The list is more useful when it is short and specific than when it is comprehensive and general, because a long checklist is signed rather than worked.

Stencil and Print Verification

The first print after a stencil change is checked before any board reaches placement, because the print is the cheapest point at which a stencil error can be found. The check covers deposit volume, alignment and the condition of the stencil underside.

A paste inspection system reports the volume per aperture, and that result is compared with a target band rather than with a single value. Where no inspection system is installed, the first board is weighed or a sample of apertures is measured. The inspection methods are described in our SPI guide.

Feeder and Placement Verification

Feeder verification means confirming the part number, the pitch and the programme position for every feeder, and it is the step most often shortened under time pressure. The consequence is a placement error that appears only where the wrong part has been placed in a position nobody inspects.

A practical control is a first article board inspected against the assembly drawing rather than against the programme, because the drawing is the independent reference. Inspecting a board against the same data that placed it finds nothing. Placement capability is discussed in our capability notes.

Profile and Thermal Verification

Where a changeover alters the thermal load on the oven, the profile is re-measured rather than assumed. Changing the product changes the mass on the conveyor, and a profile verified for the previous product does not describe the new one.

The re-measurement uses the standard attachment method for the line, and the result is compared with the reference profile for the product. Where the product is new, the first profile becomes the reference and is filed with its attachment details.

First Article Inspection Methods

First article inspection combines a visual check, X-ray of any hidden joints, and measurement of the features the process can drift on. On a fine pitch assembly that includes placement offset and fillet geometry rather than just the presence of a joint.

The inspection is documented against the drawing, with each characteristic recorded and the result compared with its requirement. A first article signed without recorded measurements gives no protection against a drift that starts on the second board.

Recording the measurement, not just the pass or fail, is what turns the first article into a baseline that later boards can be compared against.

Reaction to a Failed First Article

A failed article stops the run until the cause is identified, and the discipline that matters is changing one thing at a time. Adjusting the print and the profile together produces a result that cannot be attributed to either.

Where the failure is a placement error, the feeder and the programme are the candidates and they are checked in that order because the feeder is the more common cause. Where it is a print defect, the stencil and the support pattern come first. The effect on the line is visible in the first pass yield record.

Traceability and Records

The record of a changeover identifies the product, the revision, the machine, the operator, the time, the verification results and any deviation that was accepted. That record is what allows a defect found a week later to be traced to the job that produced it.

Where a deviation is accepted, the reason and the authority belong in the record, because an undocumented deviation becomes a precedent. The alternative is a process that drifts away from its specification one concession at a time.

Reducing Changeover Time Without Losing Control

Time is saved by preparing the next job off line: feeders loaded and verified, stencil cleaned and checked, and the support plate pre-set. None of those preparations removes the verification that happens on the machine.

What should never be shortened is the print check and the first article inspection, because those are the two steps that find a mistake while the cost is still one board. Everything else can be prepared in advance without weakening the control. The judgment applied to the finished board is discussed in our board quality guide.

The paste lot belongs on the verification list, because two lots of the same product can print differently, and a lot change during a run reintroduces a variable that was verified only at the start.

Where a machine stores more than one programme for a product, the revision is verified and the obsolete copy is removed, because selecting the wrong one produces an offset that no board inspection reveals until it is measured.

The support pin pattern is verified against the drawing rather than the machine display, because the display shows what was programmed and not what is physically present under the board.

A changeover record that includes the elapsed time is useful for planning, and it also shows which steps are being compressed when the schedule is tight.

Where a product runs for several shifts, the verification is repeated at the start of each shift rather than assumed to hold, because feeders are topped up and nozzles are changed between them.

Feeder setup is verified with a sample component rather than with the programme, because the programme confirms what the machine intends and not what is on the reel.

Where one feeder position serves two similar parts, the verification includes a visual check of the marking, since a reel that is one digit different reads identically in the machine data.

FAQ

Does every changeover need a new profile? Only where the thermal load changes, but that is most product changes, so the profile is re-measured unless the product is a proven equivalent.

How much of the first article should be recorded? At least the measured values for the characteristics the process can drift on, with the requirement beside each one.

Can offline preparation replace on-machine verification? No. Preparation saves time, while verification on the machine is what confirms the setup actually matches the product.