PCOLA-SOQ

PCOLA-SOQ ^[1] is a method to gauge the effectiveness of testing a printed circuit board assembly (PCBA) in manufacturing to ensure the PCBA is devoid of manufacturing defects.

PCBA Manufacturing Defect

The objective of PCBA manufacturing is to attach correct, working, and properly oriented components to their intended locations with connections that are formed to establish reliable mechanical and electrical connectivity. PCBA defects will evolve with technology as new component package types are adopted and assembly processes improve over time.

A PCBA manufacturing defect has been defined as “An unacceptable deviation from a norm” ^[2]. The term “unacceptable” is key; action is required to repair the defect, and/or to assure the process that allowed it to happen is improved. Defects should not be passed on.

Thus a PCBA manufacturing defect is ‘an unacceptable mounting of a part and/or an unacceptable formation of a connection (solder joint, press-fit connection, etc.).

PCBA Manufacturing Defect Universe

The PCBA manufacturing defect universe comprises all manufacturing defects potentially created in the process of assembling and attaching components to a printed circuit board (PCB). This includes defects

1. Related to components on the PCB.
2. Related to the attachments points (typically “pins”) of the components.

For every PCBA, a bill of materials (BOM), x-y locations and component orientation information are used to assemble the PCB. The defect universe is the enumeration of all the potential defects using the PCB assembly information.

The PCBA defect universe is an enumeration of ALL potential defects in the PCB assembly process and is not reduced by the defect detection capability of a manufacturing test process. The fact that a given test technology may not be capable of detecting all defects does not mean untestable defects should be ignored.

PCBA Manufacturing Test

The premise of PCBA manufacturing test is based on the concept that a defect free PCBA increases the chances of a functionally working board. Manufacturing test results strive to:

1. Pinpoint defects enabling efficient repairs to correct them. This is termed “defect isolation” and is a key contribution of Manufacturing test.
2. Provide data for quality improvements to the manufacturing process.

In most manufacturing practices, an additional functional test is performed after manufacturing test. The objective of the functional test is to ensure the board functions as it was designed and to provide additional defect coverage for those that were not covered by manufacturing test. The PCBA functional test defect universe is different from that of PCBA manufacturing test as is its test technologies. The PCBA functional test will determine whether the designed function performs as it was intended to. It may or may not isolate a failure to a particular component or connection. PCOLA/SOQ defect isolation is usually not indicative of proper PCBA function.

Generally the PCBA manufacturing test strategy complements electrical and imaging tests; minimizing redundant tests while increasing capability to detect a larger portion of the defect universe.

Electrical process test systems can be categorized into:

• Flying Probe
• Manufacturing Defect Analyzer (MDA)
• In-Circuit Test (ICT)
• IEEE 1149.1 Boundary Scan and other DFT-oriented standards

Test systems in each of the categories vary in test methodology and within each category there can be differences in each system’s test capability and throughput.

Imaging systems typically used in PCBA manufacturing are:

• Automated Optical Inspection (AOI)
• Automated X-ray Inspection (AXI)

These systems vary in the technology. AOI uses reflected light and cameras to capture images for analysis while AXI uses penetrating X-rays to inspect solder joint quality. AOI technology can be deployed at various stages of a surface mount technology (SMT) line to inspect solder paste deposition, component mounting and solder defects. X-rays can “see through” devices, for example, Ball-Grid Array devices, to inspect solder connections beneath them.

PCBA Manufacturing Test Coverage

The ideal PCBA manufacturing test covers (that is, successfully detects) any defect(s) present from the entire PCBA manufacturing test defect universe. However in practice it is almost impossible to test for 100% of the PCBA manufacturing defect universe due to electrical and imaging test limitations and tradeoffs. The electrical process test challenges can range from the lack of access(test pads) to components in parallel circuits (for example, hundreds of small capacitors in parallel across power and ground planes). Imaging test challenges can be component(s) and solder joints hidden by RF shielding, small outline surface mount device (SMD) (eg 01005) under a larger SMD (eg ball grid array (BGA)) and even hidden solder joints masked by underside heat sinks or another solder joints. Thus, each test technology has significant ‘blind spots’, and a goal is to combine technologies to more effectively eliminate untested defects.

PCBA manufacturing test can be compromised by:

• The lack of design for test (DFT) verification prior to design freeze.
• Time to market (TTM) pressures compressing the time allocated to develop, debug and verify the tests.
• Lowering overall cost of test for equipment, labor and test operations.

Previously, PCBA manufacturing used a simple formula to determine the test coverage:

$\text{Component test coverage}=\frac{\#\text{ of components tested} }{\text{Total }\#\text{ of components} }$

This formula does not account for the quality of the test for each of the components. If a test verifies only 1 pin out of 1,521 pins of a BGA, the formula will deem the component is tested ignoring potential defects to the remaining 1,520 pins of the BGA component.

The formula is not rigorous to determine:

• The quality of test of each component.
• The criticality of the defect in the defect universe.

Similarly, PCBA shorts coverage in the past was defined as the number of accessible board nodes divided by the total number of board nodes.

$\text{PCBA shorts test coverage}=\frac{\#\text{ of accessible nodes} }{\#\text{ of PCBA nodes} }$

This node-oriented measure did not take into account how nodes may have many or few connection points to device pins, and how some of these could never (in practice) be shorted. Others that could be shorted were also not tested. For example, a small-value inductor, a closed switch or jumper or a small-value resistor could have a short across its terminals, but not be tested due to resolution problems with the impedance measurement.

PCBA Manufacturing Test Effectiveness

The overall effectiveness of PCBA manufacturing test should be compared against the defect universe.

The defect universe for each component in the BOM (Bill of Materials) includes:

• Presence (the component is present)
• Correctness (it is the correct component)
• Orientation (if the component is polarized, it is not rotated by 180 degrees or for a QFP or BGA is not rotated by 90, 180 or 270 degrees)
• Live (the component is basically alive – note this is not a full functional qualification)
• Alignment (the component is properly centered, free of skews or small rotations)

The defect universe for component pin connections includes:

• Shorts (unwanted continuity to other nearby connection points)
• Opens (lack of continuity between component and the intended board connection)
• Quality (free of malformation, excess or inadequate solder, cold solder voids, etc)

The first letters of each potential defect forms the acronym PCOLA-SOQ. PCOLA-SOQ represents a more comprehensive indicator of a PCBA manufacturing test effectiveness than the historical test coverage formulas.

PCOLA-SOQ and PCBA Defect Universe

A coverage value is assigned to PCOLA-SOQ to indicate the capability of the PCBA manufacturing test process to detect the defect. It is important to note that a “test” is actually a sequence of many “experiments” that gather information. For example, an ICT may measure many individual passive components. An AOI may inspect many visible solder joints, one by one. A given experiment or inspection reveals information about a small portion of the defect universe. Each can be “graded” for defect coverage using the following criteria, based on each experiment “passing”, meaning no defect was noted in the process.

The coverage value of:

• ‘1’ is assigned for “fully tested” (complete confidence)
• ‘0.5’ is assigned for “partially tested” (some chance of error)
• ‘0’ is assigned for “untested” (no information is gained from this test)

The ideal PCBA manufacturing test(s) should have the capability to detect the entire PCBA defect universe. It should score a ‘1’ for fully testing each component in the BOM and every component connection.

The three common PCBA manufacturing test systems are AOI, AXI and ICT. Each system has its strengths and the combined defect detection capability may not cover 100% of the PCBA defect universe.

For example, while these systems can score well for PCOLA-SOQ for a resistor, some may not be able to differentiate whether the resistor is wire-wound or carbon composition. Neither will some be able to differentiate a ¼ watt resistor from a ½ watt rated resistor. Thus, the ‘Correct’ category will be “partially tested”, even though its resistance is measured. Grading of tests versus the defect universe should be done conservatively, and the criteria used should be transparent.

The PCBA defect universe can be as detailed and exact as you want it to be and PCOLA-SOQ should be used to score the effectiveness of the PCBA manufacturing test to capture defects in the PCBA defect universe.

PCOLA-SOQ Scoring

In a typical PCBA, you can find resistors, capacitors and digital ICs. A portion of the digital ICs may be connected in Boundary Scan circuit(s). How do we score PCOLA-SOQ using various PCBA Manufacturing Test technologies for these components?

Example - PCOLA Scoring for ICT

The maximum PCOLA scores for ICT can be:

A slight component misalignment (eg 10% off pad) cannot be detected electrically thus a score of “untested’ is assigned to this column.

The ‘Orientation’ is NA(not applicable) for resistor and non-polarized capacitors however a polarized capacitor requires proper ‘Orientation’. Similarly digital ICs and Boundary Scan^[3] circuits require proper IC package orientation as a 90o, 180o and 270o orientation will improperly connect the IC package pins to the pcb creating a defect.

The test coverage capability of an ICT(electrical process test) will yield a different PCOLA score for capacitors either in series or parallel circuits.

Parallel capacitors, for example bypass capacitors, pose a challenge to electrically identify individual capacitor values as the capacitance across the two nodes is the sum of all the capacitance. If the sum is a few magnitudes larger than an individual capacitor value, electrical process test will be weak in detecting “Presence”, “Correctness”, “Orientation” and “Live” thus the ICT capability to detect defects in these areas should be “untested'.

Example - PCOLA Scoring for AOI

The maximum PCOLA scores for AOI can be:

AOI test technology has no capability to detect whether a component is ‘Live’ as the test technology does not apply power to the PCBA during test/inspection.

While AOI has no capability to electrically measure the values of passive components or the characteristics of active components, it may have optical character recognition (OCR) capability to interpret the component body markings to identify the ‘Correctness’ of component. The reliability and repeatability of this capability is dependent on the availability and legibility of the component body markings.

Similarly the ‘Orientation’ of components is a visual verification of component body markings. A polarized capacitor can be marked wrongly by the component manufacturer and the AOI will ‘pass’ it if the component even it fails electrically.

Example - PCOLA Scoring for AXI

The maximum PCOLA scores for AXI can be:

Similarly AXI, like AOI, does not apply power to the PCBA during test/inspection. X-ray cannot identify the ‘Correctness’ of components. Unlike AOI, it has the capability to test/inspect for components hidden under heat sinks, RF shields or under larger components. The X-ray technology can also detect voids in solder joints which ICT and AOI cannot.

Example - SOQ Scoring for ICT

The maximum SOQ scores for ICT can be:

Electrical process test has no capability to inspect/test the quality of a solder joint, thus it will score ‘0’ for ‘Q’ in SOQ. A low value capacitor in a group of larger value capacitors in parallel will be detectable for ‘Shorts’ but not ‘Opens’ and thus the ‘untested’ score. Digital ICs (with full access but with no Boundary Scan) can be tested for shorted pins, but opens (especially on input pins) may not be detectable, depending on the quality of the test performed.

Example - SOQ Scoring for AOI

The maximum SOQ scores for AOI can be:

The strength of imaging test technology is its capability to inspect the ‘Quality’ of solder joints. The AOI is capable of inspecting for visible(as opposed to hidden) solder joints for ‘Shorts’, ‘Opens’ and ‘Quality’. If the solder joint is hidden (for example under a BGA), it loses the capability to detect any SOQ defects.

Example - SOQ Scoring for AXI

The maximum SOQ scores for AXI can be:

The advantage of AXI over AOI is the use of X-ray technology to penetrate the component body to view the profile of the hidden solder joints. The profiling of solder joints can reveal ‘Quality’ issues of solder joints such as voids and ‘no wet’. However, AXI can suffer when crowded pins on two sides of a board begin to shadow individual measurements. Some pins may need to be excluded from a test as a result.

Benefits

PCOLA-SOQ is a more rigorous method of analyzing the completeness of PCBA manufacturing test to detect all the possible faults in the defect universe. It can be used to develop a test strategy using complementary test technologies to encompass a larger portion of the defect universe.

The process of assigning PCOLA-SOQ scores forces us to think in greater detail the effectiveness of a chosen PCBA manufacturing test to cover a defect class. PCOLA-SOQ will guide the PCBA test strategy owner and the test developer to:

• Define the defect universe.
• Identify coverage gaps within the defect universe.
• Selecting and implementing complementary tests to close critical coverage gaps.
• Focus efforts to implement quality tests for critical defect classes.

PCOLA-SOQ can be as basic as you want it to be or as detailed and exact if required. A basic PCOLA-SOQ may simply use ‘1’, ‘0.5’ and ‘0’ scores as in the examples above. If a more detailed PCOLA-SOQ is desired, weights can be used to derive a value between 0 and 1 to reflect the capability of the test to detect every possible fault of a component and solder connection within a category.

References

1. ^ K. Hird, K. P. Parker, B. Follis, “Test Coverage: What does it mean when a Board Test Passes?” Proc. IEEE International Test Conference, 2002, pp. 1066-1074.
2. ^ Board Assembly TIG, iNEMI, “Test Strategy Project 1, Test Coverage Analysis” July 9, 2002, slide 11
3. ^ K.P. Parker, “Defect Coverage of Boundary-Scan Tests: What does it mean when a Boundary-Scan test passes?” Proc. IEEE International Test Conference, 2003