What is
Structural Test?
The idea behind structural test of a printed circuit board assembly (PCBA) is that if:
-
All of the correct components are present
-
All of the connections among them are correct (no short- or open- circuits)
… then the resulting PCBA will be “good”. *
How does this apply to PCBA manufacturing test? It’s very good news in manufacturing because structural tests are inexpensive to develop and perform, and they produce structural pass/fail results. In other words when testing for the presence of a component, a wiring short, or a wiring open – the result is either YES this PCBA has that specific defect or NO it does not. There are a variety of structural test technologies with automated test pattern generation that are optimized to quickly and easily run these tests against any PCBA like:
-
IEEE 1149.1 Boundary Scan (JTAG)
-
Flying Probe
-
In Circuit Test (ICT)
These test technologies are excellent both for their ability to catch manufacturing defects and their ability to resolve and express these defects in ways that are meaningful to the assembler like:
-
Wrong component at R11
-
Short between U80 pin 11 and U80 pin 12
-
Open between U10 pin 5 and U90 pin 11
*Why the asterisk? Since the test is intended to verify the structure of the assembly, it only confirms that the manufacturer faithfully reproduced the design. A board with no manufacturing defects can still fail to perform its intended functions.
Structural test is based on the assumption that the design is both functionally sound and adequately margined for the tolerances of its components. It is normal for board designers debugging prototype boards (Design Verification Test or DFT) to work out the bugs on the initial builds and them move the design into production. This gets checks off action items and gets them onto their next project. Taking a look behind the curtain let’s suppose the first few boards were “Nominal”, or built with perfectly average components in most locations. In this case once production gets to the 100’th board, you can expect some outliers that, while perfectly reproduced, do not behave as intended.
This occurs because the components each have a tolerance specification allowing a certain amount of performance variation -- over time a production line will see component values that generally drift back and forth between the specified acceptable extremes. For example a 100 Ohm +/- 10% resistor might be 100.1 ohms. If you pick the next one from the batch it might be 100.2. Several batches later you may find one that is 109.9 Ohms, and eventually if you keep measuring resistors you will almost certainly find one that is 90.1 Ohms. Sounds simple enough, but then put the resistor into a circuit such as a voltage regulator Vo = (1 + R1/R2)Vref. Then apply the regulator’s error of -1 to -2 mV/V. Then continue on for every other component in the design. Even with cutting-edge simulation software there is a point of diminishing returns in this design work and an engineering manager who draws the line somewhere.
So production needs to reproduce a design that has been vetted for some unknown X percent of all margins in the design. Unexpected behavior may come up on the batch containing the 1,000th unit or the 1,000,000th unit but eventually it will come up. This is why in manufacturing we have functional test, also known as performance test (link to functional test).
Contact CAE Integration today for a test plan optimized for your specific product.