Integrating design and test increases reliability

Integrating design and test increases reliability
By Brian Anderson, EE Times
January 12, 2004 (12:39 p.m. EST)
URL: http://www.eetimes.com/story/OEG20040112S0034

An embedded-system developer wants to be confident that a new design can be achieved quickly and efficiently. Driven by market and economic forces to rapidly bring innovative products with compelling features yet unquestioned reliability, the designer must continually weigh his or her confidence in the design and decide if it should move to the next stage in product development.

By linking measurement tools with electronic design automation (EDA) and embedded software development tools, design flaws are exposed early in the prototyping phase, resulting in shortened design-cycle time and increased reliability. Additionally, using flexible and accurate hardware and software measurements early in design offers throughput and reliability required in manufacturing test and makes the design process more efficient by reusing test software and hardware from the initial design phase through production.

Finding flaws after building a complete prototype and spinning several board revisions can become very expensive. Consequently, the designer must confirm performance of the design in the current stage before moving on with the development process. Incorporating test as early as possible in this process results in higher design confidence and allows the product to be released faster by minimizing the inefficiencies and costs of backtracking from dead ends--not just design for test, but testing while designing.

Design-for-test (DFT) languages such as Standard Test Interface Language (STIL) and Common Test Language (CTL) help reduce the time spent writing production tests by exchanging test vector information between EDA tools and manufacturing automated test equipment (ATE) systems. However they do not help with the process of gaining confidence during product development. To help the designer with product development, test tools used during development must integrate tightly with EDA and embedded-software development tools.

In an integrated design -validation environment, the test software exchanges data with EDA and embedded-software tools, controls the measurement hardware, performs signal processing and data analysis and displays the results. This arrangement offers tremendous flexibility during the design and prototyping stages.

The link to embedded-software development tools allows for the automated evaluation of changes to design parameters by automating routine tasks such as compiling and downloading code to an embedded target. For example, if the designer were implementing a finite impulse response (FIR) filter using a digital signal processor (DSP), the test software could help optimize performance of the filter by adjusting the filter coefficients based upon the measured system performance.

Suppose the new product in development that uses this filter is a TV remote control with voice recognition. The signal path for this system would have a microphone, amplifier, digital-to-analog converter, DSP and some type of RF or infrared tra nsmitter. The engineer prototyping this design with an integrated environment could use an incremental approach to implementing each section of the signal path to maximize confidence in the design before building a prototype.

Ideally, the same test software and hardware used during prototyping should be reused during the verification, validation and manufacturing stages to leverage the effort in creating the test software and configuring the test-system hardware. This requires versatile measurement hardware that is compact and economical for system prototyping on the benchtop, but also has high throughput for manufacturing test.

Although suitable for the benchtop, traditional box instruments fall short of the requirements for manufacturing since they rely on low-throughput interfaces to the host controller and lack resources for system synchronization. Conversely, automated test equipment, although fast, is impractical for the design environment due to its size, cost and lead time. It is not uncom mon for an ATE system to cost upwards of $500,000 and take three months to deliver.

The ideal hardware component for the integrated design environment is a collection of modular instrumentation based on a compact high-throughput platform such as the open PXI standard.

Equipped with an integrated design environment where test software links the simulated world with the physical world, this challenge can be met while maintaining quality and reliability.

Brian Anderson is sources and high-speed digitizers product manager at National Instruments Corp. (Austin, Texas).

See related chart