SMT Scheduling and MRP Data Accuracy

Scheduling an assembly line is a data problem before it is a planning problem. The plan is only as good as the setup times, the machine rates and the material availability recorded in the system, and those records are usually stale.

When the data is accurate, the schedule can be trusted and the line can be committed to. When it is not, the planner compensates with buffer and the factory carries work in progress that it never intended to hold.

What the Schedule Needs to Know

The minimum data set is the machine rate per product, the setup time for each change, the material availability and the yield. Each of those has an error that propagates into the plan, and the largest error usually sits in the setup time.

The rates should be measured on the actual line rather than taken from the equipment specification, because the specification excludes the losses that dominate a real shift.

Setup Times and Product Families

A setup time that is recorded as a single number for every change hides the difference between changing a feeder and changing a whole platform. The data should distinguish the elements so that the plan reflects what actually happens.

Grouping products into families with shared feeders and similar component sets reduces the average setup and makes the schedule more stable. The family definition is an engineering decision that pays for itself in the planner’s accuracy.

Bottleneck and Capacity

The plan has to be built around the constraint, which on most lines is the placement area rather than the printer or the oven. A schedule that loads every station evenly ignores the station that decides the output.

The capacity of the constraint should be recorded with its real availability, including the planned maintenance and the breaks. A rate that assumes continuous running produces a plan that is unachievable on a normal shift.

Material Availability and Kitting

A schedule is only executable if the material is on site and kitted. The system should show the kits that are complete and the ones that are waiting, because a job that starts without a complete kit will stop at the machine.

Kitting before the change is the practice that turns a scheduled setup into a short one, and it depends on the warehouse rather than on the line. The interface between the two should be defined.

Yield and Rework in the Plan

The plan should account for the yield, because a station that produces boards at a lower yield consumes more capacity per good board. Ignoring yield makes the plan optimistic in a way that is discovered at the end of the month.

Rework capacity should be planned as well, since rework consumes labour and sometimes machine time. A plan without it assumes that every board passes first time.

Work in Progress

Work in progress is the buffer that hides an inaccurate schedule and the cost that the factory pays for it. It ties up material, occupies floor space and hides defects until they are expensive to correct.

Reducing work in progress is only possible when the schedule is trusted, which is why the data accuracy work comes first. This is the same reasoning that applies to the buffer placement discussed for line balance.

Changeover Frequency

The number of changes per shift is a design decision that follows from the batch sizes, and it interacts with the schedule. Too many changes consume the capacity that the plan assumed.

Where the demand is fragmented, the alternative is to hold finished goods rather than to change the line repeatedly, and the choice depends on the value and the shelf life of the product.

Communication and Feedback

The schedule is only useful if the line reports back what actually happened, including the reason for each stop. A plan that is issued and never compared with reality cannot be improved.

The feedback should be simple enough to be completed during the shift, which in practice means a reason code rather than a narrative. The codes should match the ones used in the maintenance system so that the data can be combined.

Software and Reality

The planning system calculates what it is told, and a system that is fed optimistic data produces an optimistic plan. The remedy is a cycle of measurement and correction rather than a better algorithm.

The parameters that the system holds should be reviewed on a schedule and adjusted from measured data, with the changes recorded so that the effect can be seen. These records belong with the process data described in manufacturing processes.

Process Control and Verification

A short note on the drawing about handling, storage or packaging is often worth more than an extra decimal place on a tolerance. Where a value sits close to a process limit, the drawing should say so, since the shop can then open the process window rather than working to a nominal figure that carries no tolerance.

A stack-up that is drawn rather than described removes most of the ambiguity from a quotation, and it lets the fabricator price the board against the dielectric and copper weights that will actually be used.

Related reading: our fabrication notes, board quality and design release notes cover the same ground.

Checks Before Release

The checks that matter are the ones performed on the product rather than on a sample kept for the purpose, because a coupon that travels with the panel is the only evidence about that panel. Where a requirement can be measured, it should be measured at the point of manufacture and recorded against the board or the lot it applies to.

A parameter that is set once and never re verified drifts, and the drift is usually discovered by a defect rather than by the record.

FAQ

Why does the plan never match the output? Almost always because the setup times and yields in the system are estimates rather than measurements.

Should the schedule be frozen? A short frozen window gives the line stability, and a longer one makes the plan unresponsive, so a few days is the usual compromise.

Is a detailed schedule better than a simple one? Only if the underlying data supports the detail, since precision applied to wrong numbers makes the plan worse rather than better.

What is the first data item to fix? The setup time for each change, because it has the largest error and the largest effect on the plan.