Learning objectives
- Compare in circuit test and flying probe test on speed, fixture cost and volume fit
- Read a test coverage report and explain what it proves and what it does not
- Follow false fail and retest rules without falling into test until pass behaviour
- Localise a failure using the test printout and the schematic before sending a board to rework
Lesson
### Lesson 1: Two ways to test a board electrically
In circuit test, called ICT, presses the board onto a bed of nails fixture, a custom plate carrying spring probes for the board’s test points. Powered and unpowered tests check for shorts and opens, measure analogue component values, and confirm device presence and orientation, with some powered functional checks included. The test itself is fast, from seconds to about a minute per board, and the coverage can be high. The price is the fixture: expensive to build, slow to change, and only worth its cost on stable designs at medium and high volume. Firmware programming by in system or JTAG methods often happens at this same station.
Flying probe test makes the opposite trade. Moving probes fly to the test points under program control, so there is no fixture to build and a program can be made from CAD data in days. That makes flying probe the natural choice for prototypes and low volume work. The cost is time, minutes per board instead of seconds. Choosing between them is an economics and volume decision, not a contest with one winner, and technicians should be able to defend either choice for a given product.
### Lesson 2: Coverage, fixtures and honest failure handling
The most important document at this stage is the coverage report. It states what percentage of nodes and components the test actually exercises. A board that passes a test covering only part of its circuitry is a partly tested board, and treating it as fully proven is how low coverage escapes reach customers wearing a tested label. Test engineers and technicians read the report before they trust the green light.
Fixtures age. Worn spring pins make poor contact, create false fails, and can damage the pads they press on, so fixture maintenance is part of test quality. Retest rules exist for a reason. A failed board is diagnosed, not simply pressed again and again until it passes once and ships, because intermittent faults treated that way return as field failures. The debug workflow is a skill in itself: take the failure printout, work out from the schematic which components and nets could produce it, prove the cause, and only then send the board to rework with a real diagnosis attached. James Onsomu teaches the lean habit of acting on data rather than impressions. Test data, honestly read, is the richest data on the line, and this stage trains technicians to deserve it.
Key numbers
- ICT test time: from seconds up to about one minute per board
- Flying probe test time: minutes per board, with no fixture to build
- Program lead time for flying probe: days from CAD data, while an ICT fixture must be designed and built before the first board can test
- Coverage: read as the percentage of nodes and components actually tested, and reported with the results
Defects to spot
- Low coverage escapes: untested circuitry shipped under a tested label
- Damaged test pads: worn fixture pins pressing and scraping on the same points
- False fails: poor fixture contact mistaken for board defects
- Misdiagnosed failures: boards sent to rework without a proven cause, so good parts are replaced and the fault remains
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. A new prototype arrives in small quantity and design changes are expected. Which test method fits best?
A. ICT, because fixtures are cheap to rebuild
B. Flying probe, because it needs no fixture and programs from CAD in days
C. No test at all
D. Wave soldering
Show answer
Answer: B
2. What does a coverage report tell the technician?
A. How fast the line is running
B. The percentage of nodes and components the test actually exercises
C. The price of the fixture
D. The colour of passing boards
Show answer
Answer: B
3. A board fails, then passes on its fourth attempt with nothing changed. What is correct practice?
A. Ship it, a pass is a pass
B. Keep testing until it passes ten times
C. Treat the intermittent result as a fault to diagnose under the retest rules, not a pass to ship
D. Delete the failure records
Show answer
Answer: C
4. False fails begin rising on one ICT program. Which cause belongs high on the checklist?
A. Worn fixture pins making poor contact
B. Operators arriving early
C. Paste stored too cold
D. Labels printed in the wrong font
Show answer
Answer: A
5. Before a failed board goes to rework, what should travel with it?
A. A guess from the test operator
B. Nothing, rework will find it
C. A diagnosis proven from the failure printout and the schematic
D. A fresh set of labels
Show answer
Answer: C