Line Changeover: Design Rules and Process Limits
A line changeover is the interval between the last good board of one product and the first good board of the next, and in a high mix environment it is often the largest single component of lost capacity. The changeover is also where mistakes are introduced, because the settings that were correct for the previous product are still in place while the new one runs. Changeover control is therefore a quality activity as much as a productivity one.
What a Changeover Actually Costs
The visible cost is the time the machine is not producing, and the hidden cost is the yield lost in the first hour after the restart. A line that takes twenty minutes to change over but produces scrap for the next forty has a real changeover time of an hour, and only the first figure is usually measured. Adding the two together is what makes an honest case for preparation.
The cost also rises with product variety. Where a line runs one product for a week, a slow changeover is irrelevant; where it runs four products a day, the same changeover consumes most of the shift. That arithmetic is what justifies investment in feeders, trolleys and pre staged tooling. The case is straightforward once the lost output is valued at the contribution margin of the product rather than at zero.
Changeover as a Discipline, Not a Technique
The technique is familiar: convert internal work into external work so that as much as possible is prepared while the machine is still running. The discipline is harder, because it requires that the preparation is complete and correct rather than merely started. That distinction is what separates a changeover that improves with practice from one that stays unpredictable.
That means a kit for the next product is assembled, checked and staged before the current run ends. A feeder that is loaded but not verified, or a stencil that is fetched but not inspected, moves the fault into the middle of the changeover instead of removing it from it. Our production process flow notes describe where those checks belong. The kit should also contain the consumables, labels and documentation that the run itself will need.
Tooling and Feeder Preparation
Feeders are the usual bottleneck because they carry the most setup information. A trolley that holds a complete set of loaded feeders for the next product can be exchanged in minutes, provided the feeder positions and the reel data have been verified before the trolley is wheeled into place.
Stencils deserve the same treatment. A stencil should be identified, inspected and staged with its squeegee and its paste, and it should be checked against the product revision rather than against the product name. A change of aperture that was never released to the shop is a common cause of a first article failure. Storing each stencil with its product and revision marked on the frame removes most of that risk.
Programme and Recipe Verification
The machine programme carries the placement data, the fiducial type and the conveyor width, and all of it has to match the product. Where a recipe is called up from a library rather than entered, the risk is not the typing but the version, because an old recipe can appear perfectly valid.
Verifying the recipe against the released documentation takes a minute and prevents a run of misplaced parts. That check is worth doing on the first article of every changeover, not only on the first time a product is run. Where the recipe is generated from CAD data, the check confirms that the correct revision was exported.
Material, Paste and Stencil Handling
Paste handling follows the changeover closely, because a jar that was opened for the previous product may not be suitable for the next. The working life of the paste is continuous, so a changeover that takes an hour consumes an hour of the paste’s life whether or not it was printing.
Material should be staged by kit rather than collected during the change, and every reel should be scanned against the bill of materials as it goes onto the machine. Scanning is slower than trusting the label and much faster than reworking a finished batch. It also creates the traceability record for the batch without additional paperwork.
Setup Time Reduction Without Risk
The way to reduce setup time is to remove the searching, the walking and the waiting rather than the checks. Preparation is external work, and moving work outside the stopped machine is the whole principle. A shadow board, a trolley in the right place and a kit that is complete at the start of the changeover remove most of the wasted time without touching the verification steps.
Standardising hardware helps as well. Common nozzle types, common feeder models and common stencil frames reduce the number of items that have to be found and adjusted, and they make it possible for an operator to prepare a kit without specialist knowledge.
First Article and Restart Control
The first article is what confirms that the changeover worked, and it should cover the print, the placement and the reflow result rather than only one of them. A first article that passes the print and is never checked after reflow has verified only part of the process. Our SPI and AOI notes describe the measurements available at each stage.
The restart should be conditional on the first article passing, with a defined disposition for parts produced before it was checked. Where that rule is not written down, the temptation is to run the line while the result is being evaluated, and the parts produced during that window are the ones that fail later.
Measuring and Improving Changeover
Measurement is the starting point, and the useful definition is the time from the last good board to the first verified good board. Recording it per product shows which changeovers are the expensive ones and whether the improvement work is having any effect.
Recording the reasons for delay is as important as the duration. A chart of causes usually shows that most of the time is lost to a small number of recurring problems, such as a missing item or a programme that had to be corrected, and those are the ones worth fixing first. Publishing the chart at the line makes the improvement visible to the people who have to deliver it.
Documentation and Training
A changeover procedure that lives in a binder is not a procedure. The steps should be visible at the line, with a checklist that an operator can complete and sign, and the order of the steps should match the way the work is actually done. A checklist that describes an idealised sequence nobody follows will be signed without being read.
Training then becomes a matter of practice rather than explanation. Our fabrication notes checklist and our judging PCB quality guidance describe how the records and the acceptance decisions are kept together for each run.
FAQ
What is a realistic changeover target? It depends on the machine and the product, but a well prepared line with a pre loaded trolley can reach ten to twenty minutes for a similar product. A change of stencil, paste and programme on an unfamiliar product will take longer.
Does faster changeover reduce quality? Only if the checks are removed rather than the wasted movement. The verification steps are what make a fast changeover safe, and they are usually a small part of the total time.
How does gopcb manage changeovers? We kit and stage material before the run ends, verify the programme and the stencil against the released revision, scan every reel against the bill of materials, and hold the restart until the first article has been measured and accepted.
