SMD Changeover Verification: Setup Checks Before the First Board

An SMD changeover is the interval between the last board of one product and the first acceptable board of the next, and it is the point in a shift where most errors are introduced. Feeders are swapped, programs are loaded, the stencil is changed and a different thermal recipe may be selected, all within a few minutes and often without a formal check on any of it.

The verification described here treats an SMD changeover as a sequence of independent confirmations rather than as a single event. Each confirmation has an owner, a record and a pass criterion, which is what makes the interval short and repeatable instead of short and lucky.

What a Changeover Changes

Four things change: the parts presented to the machine, the programs that drive placement, the printing tooling, and the thermal recipe. A part change without a program change, or a program change without a part change, is a mismatch that produces a systematic defect across the whole run rather than a scatter of random ones.

Mismatches survive because the machine still runs. Placement continues, paste is still printed and the oven still heats, so nothing stops the line. Only a check against released documentation reveals the mismatch, and only a first article inspection reveals its effect. Neither check takes long when the documentation and the inspection criteria are already at the line.

Feeder Setup and Part Verification

A feeder setup begins with the load list and ends with a comparison of every reel position against the program. Manual confirmation by two people is slower and less reliable than a barcode scan, because the scan compares the value held by the machine with the value in the file and reports a difference rather than relying on attention. The same scan confirms the reel count and the remaining quantity, which catches a part that has been loaded in two positions at once.

The mechanical side matters as much as the data. A feeder that is seated but not locked, a tape that is not threaded to the pickup point and a splice thicker than the carrier tape all cause faults later in the run. Feeding two or three components before the run confirms the mechanical state.

Stencil and Paste Changeover

A stencil change brings its own checks: part number, revision, thickness and mesh tension. A stencil with the correct part number but an earlier revision prints apertures that no longer match the pad layout, and the difference is often small enough to survive a visual look.

Paste is usually a shared material, so the check is on its state rather than its identity. Paste on the stencil at the start of a new run should be at working viscosity, within its open time and at the correct temperature. The printing side of the same variables is covered in the notes on print speed and pressure.

Reflow Profile and Product Recipe

The reflow profile is selected rather than created at a changeover, and the selection is a match between the loaded recipe and the product. Recipe name, revision and the number of zones the profile uses should all be confirmed, because a profile written for a thicker board will under-heat joints on a thin one.

Confirmation is cheap when the oven recalls recipes automatically and expensive when zones are set by hand. Where zones are set by hand, the setpoints and the conveyor speed are recorded and signed, and the profile is confirmed against the thermal mass of the product. Conveyor guidance on the same equipment is described in the notes on conveyor edge clearance.

First Article Inspection

The first article is the only sample that can be inspected before the run proceeds, so it carries more weight than its size suggests. Placement position, rotation and paste volume are measured on the first few boards and compared with the limits in the program.

An inspection that consists of looking at the board under a lamp is not a first article. The measurements that matter are numeric: offset in both axes, rotation in degrees, and volume as a percentage of nominal. A print inspection system that reports those numbers continuously is described in the notes on print feedback.

Line Clearance and Old Parts

Line clearance confirms that nothing from the previous product remains within reach of the machine. Reels, sticks, trays and loose components belonging to the old product are removed and returned to stores, and the areas that are hard to see, such as the underside of feeders and the return channel, are checked as well.

The check is documented with the position of every remaining feeder rather than with a statement that the line is clear. A signature that says the line was cleared is a claim, while a map showing which feeder positions hold which part number is evidence.

Program and Recipe Version Control

Programs and recipes change between runs for legitimate reasons, and the changeover is the moment when the current revision is confirmed rather than assumed. The released revision, its date and the person who released it are recorded at the machine, so a later investigation can establish which version produced a given batch.

Machines that hold several revisions of the same program make this check essential. If the operator selects a program by name, the revision number should appear on the same screen, because a program chosen from a list of similar names is a common way for an old revision to be used. Where programs are held on a central server, the download event is itself a record, and its timestamp ties a batch to a revision without anyone writing it down.

Verification Records

The record of a changeover has to be short enough to be completed and specific enough to be useful. The fields that carry the most value are the product, the machine, the program revision, the feeder map, the stencil revision, the paste lot and the first article results.

Those fields allow a defect found later in the run to be attributed to the changeover or excluded from it. Without them, the investigation starts from the assumption that the changeover was correct, which is the assumption that wastes the most time.

Common Failure Patterns

The recurring failures are predictable. A feeder holding the correct part in the wrong position, a stencil revision not recorded, paste left on the stencil from the previous product, and a profile selected by name alone are the four that appear most often.

Each is caught by a check already listed above, which is why the value of the procedure lies not in its novelty but in whether every step is performed at every changeover, including those made at the end of a shift when the line is running late.

FAQ

Does every changeover need a first article inspection? Yes where the product has fine pitch parts or tight placement tolerance. The measurements on the first board are the only evidence taken before the run proceeds.

Is barcode verification of feeders worth the time? It replaces a manual comparison that fails silently. The scan compares the reel against the program and reports a difference instead of relying on attention.

How long should verification take? The checks are the same at every changeover. The time comes from the number of products and the state of the tooling, not from the number of steps.