Functional test: Back to basics
Antonio Grassino, President, Seica, Strambino, Italy -- Test & Measurement World, 10/1/2004
Despite the proliferation of electrical test and inspection techniques, functional test remains necessary to ensure that a product will work in its final environment. In fact, functional test is often mandatory for military, aerospace, and automotive applications where product safety must be guaranteed. The analog, digital, memory, RF, and power circuits in a product often require different functional test strategies. A thorough functional test requires that a manufacturer test a very large number of critical paths and also perform structural verification (absence of hard faults) to catch escapes from previous test stages. When performing these tests, the manufacturer must continuously apply a large number of analog/digital stimuli to the unit under test (UUT) while monitoring their responses.
Functional test can be implemented in many forms, each offering advantages and disadvantages in terms of cost, time, effectiveness, and maintainability. Here are four basic categories: mock-up test subsystem, test bench, specialized test equipment, and automatic test equipment.
Mock-up test subsystemConceptually, the simplest way to verify whether a unit functions properly is to plug it into a "golden mock-up" system or subsystem and verify that all works fine. If everything works, then there is a high level of assurance that the unit is good. If the unit fails, the technician will interact with the subsystem, trying to identify the cause of the failure and guide the repair operations. This plug-and-play approach suffers from a number of disadvantages and is rarely effective, though it is sometimes deployed as a complement to other test strategies.
One disadvantage of such a system is its cost: A golden subsystem might often be more expensive than a traditional test platform, especially if the latter is general-purpose and can be used on multiple applications. Furthermore, maintaining this subsystem in "golden reference" conditions can be complex, time consuming, and costly. A centralized depot repair facility can be quickly filled with a proliferation of mock-up subsystems, each requiring specific documentation and training, operating instructions, and maintenance.
Also, it is not sufficient to simply plug the UUT into the system: An adequate operating sequence has to be executed to ensure the UUT operates correctly, or to diagnose why not. Developing dedicated test sequences can be costly and time-consuming, and skilled technicians are required to execute them.
Finally, even if special adaptations are made, debugging a unit on its system is cumbersome and unpractical. The very limited control on the flow of operations and the lack of diagnostic tools will quickly make this approach economically unacceptable.
Test benchA test bench is a conventional test environment including stimuli/responses and an interface to the UUT, with sequence and control driven by a specific test procedure. Stimuli and responses are usually provided by standard power supplies and lab instrumentation, dedicated switches, loads, and sometimes custom electronics. Fixturing is the critical part of the assembly, as it provides the correct signal routing and connection to the UUT. For most applications, a manufacturer must heavily customize a fixture and set it up manually. Test sequence and control is generally manual, sometimes assisted by a PC, and is defined by a written protocol or procedure.
Test benches geared to a specific product are generally low in cost and require a limited infrastructure. But they lack the flexibility to deal with multiple products, and even for a specific product, they can be inadequate when a large number of stimuli/responses are required.
Test benches are often found in engineering departments, where available instruments can be easily connected together, knowledge is at hand, and no formal process is required. With few exceptions, test benches for high-performance products are not adequate for production test.
Specialized test equipmentThe third option, specialized test equipment (STE), can automate the operations of a test bench. In an STE system, a computer generally controls a collection of programmable instruments over a dedicated bus (e.g., IEEE, VXI, PXI, or PCI).
Speed, performance, cost, and other factors will affect the choice of the instrument bus and overall architecture. Instruments and general-purpose resources are stacked into one or more vertical cabinets and are routed to the UUT.
Because routing and connections are usually automated and controlled by software, the internal wiring to the receiver can become complex. Digital resources (channels) are usually on a dedicated rack, while a separate rack contains a switching matrix to connect and distribute analog instrumentation. When both analog and digital channels are needed, jumpers are provided on the fixture.
To optimize cost, space, and flexibility, the configuration is often focused on a specific project or program. Thus, each new project requires new STE. Thanks to automation, setup time, test time, and overall operations are much quicker and simpler than with a manual test bench.
Test-program generation itself is not necessarily simpler, but the documentation required will be reduced. STE
 |
| Figure 1. This ATE system can perform analog, digital, and high-frequency functional tests. |
STE also has drawbacks, the most notable being overall cost, including capital costs, operating costs, and program-development costs. Capital costs include engineering, material, manufacturing, test, documentation, and depreciation costs. Operational costs include fixtures, maintenance and spares, facilities, indirect materials and consumables, labor, and overhead. Finally, for each type of unit,the costs for test-program development and debug should be computed and added.
Automatic test equipmentGeneral-purpose automatic test equipment (ATE) represents the most flexible approach for satisfying functional test requirements for a wide variety of products and programs (Figure 1). System integration, flexible signal routing, value-added hardware and software, test-oriented languages, and graphical user interfaces are a few key differentiators between ATE and STE. Though software and hardware plug-and-play standards are today proliferating—and "open-system" is becoming a buzzword—you should not underestimate the value of proprietary integrated solutions.
The advantage of being commercially available gives ATE the benefit of its manufacturer's years of experience and nonrecurring engineering (NRE) investment. This is particularly valuable when the ATE supplier adopts new technologies while maintaining existing features and techniques. This is of utmost importance for military and aerospace products that have a long life and where old and new assemblies coexist and will continue to be tested.
Digital channels for parallel test are one of the major elements of ATE. Usually, ATE systems use proprietary architecture, as they are designed to address a wide range of test requirements. Speed, control, depth of data, timing flexibility, and voltage-level excursion are some of the characteristics to look at to understand how easy it would be to bend the system to comply with test requirements.
Serial digital test, with a variety of protocols, is usually provided by dedicated instrumentation integrated into the system's architecture. That would also be the case for IEEE 1149.2 or JTAG/boundary-scan test technology, which can then be integrated into a comprehensive test environment.
Like with the STE architecture, commercial instrumentation is integrated into an ATE system's architecture to provide analog test capabilities (Figure 2). Besides offering full instrument integration, ATE solves signal routing and connection problems.
The backplane of most ATE systems includes an analog bus that allows direct routing of instrumentation to
 |
| Figure 2. The core architecture of this ATE system is compatible with PXI, VXI, and IEEE 488 instruments. |
Modularity is designed into the ATE so its general-purpose nature can be fully exploited from one project to another: same system, same software, same training and documentation, same operations.
Whether deployed for engineering or production test, ATE will be part of a process, and it includes its own structured process for optimal use. Test-program generation employs links to a CAE database.
Program-generation, whether manual or simulation-driven, is usually well-structured to allow links to external program sources, concurrent test generation, graphical programming, seamless modifications, self-documentation, and a comprehensive link with debug. Debug and run-time features include facilities such as stop on fail, looping, conditional branches, changes in real time, analog and digital internal probing, and other facilities that simplify the work of the programmer and the operator.
| Author Information |
| Antonio Grassino has been involved in the board test industry for more than 30 years, working in R&D, technical support, application, and sales. He is a founder of Seica, of which he is currently president and sales director. |
