Learning objectives
- Run a board powered up in a product like fixture and exercise its real functions
- Apply test limits taken only from the product specification
- Log every result against the board serial number for traceability
- Separate fixture contact faults from genuine board defects on intermittent failures
Lesson
Electrical test proves a board is built correctly. Functional test proves it works. At this stage the board is powered in a fixture or rack that imitates its life in the product, and it is exercised as a product: power rails are measured under load, communication interfaces are driven, sensors and displays are worked, firmware functions are run, and calibration is performed where the product requires it. Custom fixtures, power supplies, loads and data acquisition hardware run under a software test sequencer, and every result is logged against the serial number of the individual board.
Limits are the first discipline. Every pass and fail boundary at this station comes from the product specification. Nothing is invented at the bench, adjusted from memory, or borrowed from a similar product, because a wrong limit manufactures two kinds of damage at once: good boards failed and scrapped or reworked for nothing, and faulty boards passed into the field wearing a green label. Fixture and software revisions are controlled for the same reason, and calibration status is checked before results are trusted.
Intermittent failures are the teaching core of this stage. A board that fails one run in five may carry a genuine marginal defect, or it may be innocent and the fixture contacts are tired. Where the product is a complete unit rather than a bare board, the same discipline extends into system test, with the finished product run through its operating modes, its accessories connected and its user facing functions proven before it is allowed anywhere near a packing bench. The trained technician separates the two methodically: reseat the board, inspect and clean contacts, run the board on a second fixture if one exists, and watch the marginal readings themselves. A reading that sits close to its limit, pass after pass, is data speaking early. Waving it through because it is technically a pass is how drift becomes a field failure trend. James Onsomu teaches in his lean writing that teams should analyse performance data and use it to drive improvement. Functional test data, kept by serial number and read honestly, shows which products are drifting, which fixtures are lying, and where the process needs attention next. That closes the loop between testing a board and improving the line that built it.
Key numbers
- Test limits: taken only from the current product specification, with no bench invented values
- Data retention: pass and fail results logged against the serial number for 100 percent of boards
- Fixture and software: revision controlled and calibration in date before results are trusted
- Marginal readings: recorded and trended, never waved through as unimportant
Defects to spot
- Intermittent failures: tired fixture contacts imitating board defects, or real marginal defects imitating fixture faults
- Marginal results ignored: readings drifting toward a limit that nobody trends
- Wrong firmware or limits loaded: revision errors that fail good boards and pass faulty ones
- Missing traceability: results not logged against the serial number
Stage check, self test
Answer each question, then open the answer to check yourself. Passing standard: 80 percent, at least 4 of 5 correct.
1. Where must functional test limits come from?
A. The product specification
B. The operator’s experience with similar boards
C. A whiteboard in the test area
D. The customer’s receptionist
Show answer
Answer: A
2. A board fails one run in five at functional test. What is the trained response?
A. Ship it, it passed the other four
B. Methodically separate fixture contact faults from a genuine marginal board defect
C. Raise the test limits until it passes every time
D. Remove the test from the route
Show answer
Answer: B
3. Why are results logged against the serial number of each board?
A. To fill storage space
B. To slow the line down
C. For traceability, so any board’s full test history can be proven later
D. To help the label printer
Show answer
Answer: C
4. A voltage reading passes but sits very close to its limit on board after board. What should happen?
A. Nothing, a pass is a pass
B. The reading is recorded and trended as early warning data
C. The limit is quietly moved
D. The fixture is unplugged
Show answer
Answer: B
5. Wrong firmware is loaded at the functional test station. What is the likely result?
A. Nothing changes
B. Boards test faster
C. The fixture lasts longer
D. Good boards fail and faulty boards can pass, because the board is judged against the wrong behaviour
Show answer
Answer: D