Component Loss on SMT Lines: 5 Accounting Rules
Component loss is the difference between the parts issued to a placement line and the parts that leave it on shipped boards. On a dense product the number is small in percentage terms and large in absolute terms, because a board can carry a thousand chips and the line runs a thousand boards a week.
Measuring component loss matters for two reasons. The first is cost, because the loss is real material that was paid for and scrapped. The second is diagnosis, because a line that loses more than its baseline is usually telling you about a nozzle, a feeder or a tape that will soon cause a defect rather than a variance.
What Counts as Component Loss
The definition has to be written before the number is measured. Parts that are scrapped with a defective board are a loss, and so are parts dropped inside the machine, parts consumed by a first article, and parts discarded when a reel is left with a short tail. Parts returned to the store are not a loss.
Material variance is the accounting view of the same figure, and it is usually reported as a cost rather than as a count. Where the two are kept apart, the shop sees a financial number once a month and a placement problem once a quarter, and the connection between them is never made.
The Sources on a Placement Line
Feeder loss comes first. Every reel change consumes a leader and a trailer, every splice adds a joint that can mis-pick, and every setup consumes a few parts while the feeder is taught. A line with forty feeders and two changeovers a day generates a steady, predictable loss before any defect occurs.
Pick errors come second. A part that is picked badly is usually blown off the nozzle or dropped on the way to the board, and it lands on the machine floor rather than on a board. Damage at the nozzle is a third source: a cracked capacitor is discarded by the operator and never reaches the counter.
Machine Counters as a Measurement Tool
Modern placement machines count pick attempts and successful placements per feeder, and they log the errors between them. That log is the cheapest measurement available, and it separates a feeder that is losing parts from a program that is placing the wrong number.
The counter has limits. It cannot see a part that was picked and later discarded by an operator, and it cannot see components lost when a panel is scrapped. Used alongside a physical count it is a strong diagnostic, and the difference between the two numbers is the manual loss that the machine cannot explain.
Weighing Reels and Partial Reels
Counting by mass is the standard way to fix a physical number. The component mass is known from the data sheet, so a reel can be counted on a laboratory balance and the count converted. A reel of five thousand 0402 chips weighs a few grams, which means the balance has to resolve a hundredth of a gram.
The method works best for partial reels, where the question is how many parts are left. It is less useful for a full reel, where the nominal count is reliable, and it is not a substitute for the counter in a machine. Both tools answer the same question from different directions, and the feeder inventory record is where the two come together.
Reconciliation per Work Order
The clean way to measure component loss is per work order. Issue the reels, record the counts, run the batch, and count what comes back. Add the parts on good boards, the parts on scrapped boards and the parts returned, and compare with the parts issued. The remainder is the loss for that order.
A monthly total hides the pattern, while a per order figure shows which product and which feeder are responsible. The comparison is worth doing even when the number is small, because the trend is the early warning. A program whose loss has doubled is a program with a feeder that is about to fail.
Loose Parts Inside the Machine
A dropped component does not disappear. It lands inside the machine, on the conveyor or on the floor, and a loose metal part that reaches a board can short two pads. Loose parts are therefore both a loss and a quality risk, and the risk is worse than the cost.
The control is a cleaning routine with a fixed interval and a tray under the placement area. The tray is also a measurement: the parts collected in it over a week are the physical evidence of the loss the counter reported. Where the tray fills quickly, the nozzle and the pickup reliability should be checked before the next batch.
Setting a Target and Reading Deviations
Set the target from the process rather than from an ideal. Count the loss on a stable product over several batches, and use the average as the baseline. Then set a limit, such as a tenth of a percent above the baseline, at which the loss is investigated rather than merely recorded.
Read the deviations as symptoms. A rise that follows a reel change is a tape or splice problem, a rise that appears after a nozzle change is a pickup problem, and a rise that appears with a new product is a program or a feeder setup problem. The feeder setup verification step is where most of those causes can be removed before production starts.
Reducing Loss at the Feeder Setup
Most avoidable loss happens before the first board is placed. A feeder that is taught on the wrong tape, a nozzle that is fitted without a teach, and a program that starts with a trial placement all consume parts. A prepared setup removes them without any change to the machine.
The practical measures are a setup checklist, a dedicated partial reel store for first articles, and a rule that the trial placement is done on a scrap panel rather than on the first production board. The feeder loss record should show the improvement within a few batches, and if it does not, the loss is coming from somewhere else.
Splicing, Tails and Short Reels
The end of a reel is a routine source of loss. A splice that is made too close to the pickup point will mis-pick, and a tail that is too short to reach the feeder leaves a handful of parts that cannot be used. Both are small numbers that repeat on every reel change.
The answer is a written standard for splice placement and tail length, applied by everyone. The standard should also say what happens to a short tail: whether it is returned to the store, consumed on a prototype or scrapped. Without that rule the parts accumulate in a drawer and appear later as an unexplained variance.
Reporting the Number So It Gets Used
Report component loss beside the first pass yield rather than in a separate material report. When the two numbers appear on the same page, a rise in loss and a fall in yield are read together, and the connection between a feeder and a defect becomes visible to the people who can act on it.
The report should be short and per product. One line per product with the issued count, the placed count and the loss percentage is enough, and it should be issued weekly rather than monthly so that the trend can be acted on within the same production cycle.
FAQ
Is a small component loss normal? Yes. A placement line always loses some material through reel changes, splices and setup, and the useful question is whether the loss is stable. A stable loss is a cost that can be budgeted, while a rising loss is a process that needs attention.
Should component loss include parts on scrapped boards? It should, because those parts have left the material store and will not be shipped. Separating them from placement loss is useful for diagnosis, but both belong in the total that is compared with the parts issued.
How accurate is a machine counter for component loss? It is accurate for pick attempts and placements, and blind to anything an operator removes by hand. Treat it as one of two measurements, and use the physical count per work order to close the gap.
