How to Improve First Pass Yield (FPY) in SMT Manufacturing — A Practical Guide

To improve first pass yield in SMT manufacturing, start with a reliable first-attempt count. Find the largest source of loss, connect failed boards to their production history, and verify a corrective action against comparable builds.

This guide covers the FPY formula, a worked example, retest rules and a practical investigation workflow for electronics manufacturers. It applies to a single inspection or test station, or to a clearly defined production route.

Updated 9 November 2026 · Formula, counting rules and investigation steps reviewed.

What is first pass yield?

First pass yield (FPY) is the percentage of units that meet the requirements of a defined process on their first attempt, without repair, rework or retesting. A board that passes after a failed attempt contributes to final yield, but does not become a first-pass success.

In SMT, define the boundary before reporting the number. “AOI FPY” describes first-attempt results at automated optical inspection. A line-level measure might cover assembly through final functional test. Those numbers answer different questions and should be labelled separately.

The definition follows ASQ’s quality glossary. The SMT counting policy and examples below explain how to apply it to production data.

FPY calculation: a worked SMT example

Suppose a completed production lot contains 1,000 boards. Of these, 874 pass the defined route on their first attempt. Another 86 pass after repair and retest, and 40 are scrapped. These are illustrative figures, not customer results.

The gap between FPY and final yield is 8.6 percentage points. It shows how much output was recovered after the initial failure. Reporting only the 96% final yield would conceal the extra work.

Do not equate every first-pass failure with one rework event. A board can be scrapped, retested without repair, or repaired more than once. Calculate rework cost from recorded labour, material and test activity, and report scrap separately.

How to count retests, false failures and unfinished boards

• Retest after a failure: preserve the original result and attempt sequence. A later pass improves recovery or final yield, not the original first-pass result.
• Confirmed false failure: investigate fixture contact, calibration, test software and limits. If your policy allows validated exclusions, retain the raw FPY and report the adjusted figure with the reason and excluded count.
• Duplicate records: remove duplicated messages using a reliable event identifier. Do not mistake repeated data ingestion for another physical test attempt.
• Required test sequence: several different tests can form one normal route. Define the required sequence so a planned second test is not confused with repeating a failed test.
• Work in progress: use a completed lot for a reconciled result. For live reporting, show pending boards and mark FPY as provisional. Do not compare today’s completions with unrelated boards started today.
• Panels and individual boards: choose one counting unit and map panel positions to board serials. Mixing panels in the denominator with boards in the numerator produces an invalid rate.

A useful report shows first-attempt passes, first-attempt failures, pending units, retests, recovered units and scrap. If the first-attempt count changes after a data correction, record the reason so shifts remain comparable.

FPY, final yield and rolled throughput yield

Station FPY measures first-attempt success at one defined step. Final yield includes accepted units recovered through repair or rework. Rolled throughput yield (RTY) combines first-pass yields across successive steps.

Minitab explains this multiplication method. Apply it to compatible steps and populations. Multiplying dashboard averages from different products, periods or routes can give a misleading result. Where serial-level history is available, also count the boards that completed the whole route without a first-attempt failure.

A high FPY does not by itself establish process stability or capability. For measured characteristics with specification limits, use the appropriate control-chart and process capability analysis alongside yield.

Where to investigate SMT yield loss

Use the failure pattern to decide which records to inspect. The station detecting a defect may be different from the process that introduced it. The following are investigation leads, not confirmed causes.

• Solder paste printing: compare solder paste inspection (SPI) measurements, pad locations, stencil cleaning, paste handling and printer setup for passing and failing boards.
• Component placement: check placement offsets, feeder and nozzle records, component orientation and material changes.
• Reflow: review the validated profile, recipe changes and available process measurements. An oven setpoint alone does not establish the thermal history of each board.
• Materials: compare component and PCB lots while accounting for the products, lines and time periods on which they were used.
• Inspection and test: examine AOI classifications, in-circuit test (ICT) failures, functional circuit test (FCT) results, fixture maintenance and test-program revisions.

A missing data connection is an investigation limitation, not proof of a manufacturing defect. Start with the evidence available and make missing coverage visible.

A six-step workflow to improve first pass yield

1. Establish a trustworthy baseline

Document the process boundary, counting unit, required route and retest policy. Reconcile a completed lot to its inspection, test, repair and scrap records. Show the board count beside the percentage.

2. Separate changes in product mix from changes in performance

Compare the same product and revision before combining results. Then examine line, station, shift, material lot and test-program version. A factory-wide average can fall simply because a more complex product accounts for more of the output.

3. Prioritise a specific failure pattern

Build a Pareto view of first-attempt failures by defect family, test step or component location. A board can have several defect codes; distinguish the number of defective boards from the number of recorded defects. Choose a problem using frequency, business impact and customer risk.

4. Check the measurement system and connect the board history

Before changing the production process, check whether the failure is reproducible and whether test conditions changed. Use test data analytics to examine first attempts and retests, then compare the affected serials with relevant upstream records.

5. Test a cause and document the corrective action

Compare passing and failing boards under similar conditions. Account for product, lot and time effects before treating a correlation as a cause. Use engineering inspection and an approved controlled trial to validate the hypothesis. Record the action, owner, affected equipment and deployment time.

6. Confirm that the improvement holds

Compare equivalent builds after the action using the same FPY definition. Track the targeted defect, sample size, retest rate, scrap and downstream escapes. Use a suitable control chart to distinguish sustained change from routine variation. An easier test limit or a shift in product mix should not be mistaken for process improvement.

What production data should you connect?

Start with the fields needed for the selected investigation. Manufacturing data integration should preserve traceability and event order, not just collect more files.

• Identity and route: board serial, panel position, product revision, order, lot and required process steps.
• Event detail: station, timestamp and time zone, test or inspection step, attempt number, result, measured value and limits.
• Process context: available SPI, placement and reflow records, component lot, recipe and software versions.
• Disposition: defect classification, inspection outcome, repair action, retest result and scrap reason.

Validate serial mappings, clock alignment, units of measurement and record completeness. If records can only be matched by lot or time window, label that limitation. Report the proportion of boards successfully matched; a partial dataset can bias the comparison.

How QualityLine supports FPY investigations

The QualityLine platform brings production and test data into a connected analytics workflow. SMT analytics and root cause analysis help engineers investigate relationships across available station data. Engineers still validate the suspected cause and corrective action.

For the commercial use case, see QualityLine’s first pass yield improvement solution. Explore customer stories for examples of how manufacturers use the platform. Integration scope and expected outcomes should be assessed against your systems and data quality.

Frequently asked questions

What is a good FPY for an SMT line?

There is no single target that fits every product and route. Set a target using a stable baseline for the same board family, revision, test coverage and process boundary. Review it alongside customer requirements, scrap and defect escapes.

Does a board that passes retest count toward FPY?

Not when it failed its first required attempt. Keep the original failure in the FPY record and count the recovered board in final yield. Document any validated false-failure adjustment separately.

Is FPY the same as final yield?

No. FPY counts first-attempt successes. Final yield also includes units accepted after repair or rework. Reporting both makes recovery work visible.

What should we do first when FPY drops?

Check the counting rules, data completeness and recent product or test changes. Then isolate a specific failure pattern, validate the measurement system and investigate comparable passing and failing boards.

Your next production review

1. Choose one product, route and completed lot.
2. Reconcile first-attempt passes, failures, recovery and scrap.
3. Select one recurring failure pattern and check measurement reliability.
4. Assign an investigation owner and record the evidence needed to test a cause.
5. Review the next comparable builds after the corrective action.