Payoffs from merging design and test

By Larry Maloney, Contributing Editor -- Test & Measurement World, 2/1/2011 12:00:00 AM

JAMES TUNG The MathWorks Fellow and Chief Strategist The MathWorks Natick, MA Jim Tung has nearly 30 years of experience in the technical computing software field. A 22-year veteran of The MathWorks, he held the positions of VP of marketing and VP of business development before assuming his current role focusing on business and technology strategy and analysis. Tung previously held marketing and sales-management positions at Lotus Development and Keithley DAS, a pioneering manufacturer of PC-based data-acquisition systems. Tung earned a bachelor's degree from Harvard University. Contributing editor Larry Maloney conducted a phone interview with Jim Tung on the use of software-analysis and modeling tools for test applications. James Tung answers more questions on new software tools that help engineers solve complex test challenges in the continuation of this interview.   To read past "Viewpoint" columns, go to www.tmworld.com/viewpoint. Read more from the February 2011 issue.

Q: How can software-analysis and modeling tools help test engineers cope with time-to-market challenges?
A: Matlab helps engineers do more sophisticated analysis of test data much more quickly by automating engineering tasks. Combine that benefit with our Simulink modeling tools, which help engineers find errors in a product or system design very early in development. If those errors are allowed to carry over to the integration and testing phases, the result is a lot more engineering rework, which can hurt time-to-market performance. Simulating system behavior under different conditions and parameters results in a more mature, better-vetted design at the testing stage.

Q: How do test engineers work with Matlab and Simulink in a typical application?
A: Matlab is used heavily in R&D work to analyze data and ideas on a design. Simulink takes that data and creates a model of, say, a control design or signal-processing system. In many cases, Matlab is then used again to automate a set of Simulink simulations to explore design alternatives, do Monte Carlo experiments, or assess different parameters affecting a design.

Q: What are the principal Matlab toolboxes that test engineers find most useful?
A: There are several that help to manage a test system, including the Data Acquisition Toolbox, the Instrument Control Toolbox, and the Image Acquisition Toolbox. All are very useful for acquiring data and outputting stimuli. The newest release of the Data Acquisition Toolbox also enables engineers to use National Instruments' CompactDAQ data-acquisition hardware directly from Matlab. On the Simulink side, both design and test engineers use our tools for modeling physical components, such as mechanical and hydraulic systems, and for generating C code from those models to run dynamic tests like hardware-in-the-loop simulations.

Q: What has been the response to the new toolbox focusing on vehicle networks?
A: Quite good, especially with the flurry of new hybrid and electronic vehicle systems that are being developed. This toolbox connects Matlab and Simulink to a vehicle's CAN (controller area network). This eliminates the need for additional connectivity tools and creates a powerful environment for streamlining test, verification, and analysis of functions linked to CAN bus, such as powertrain, lights, body electronics, door locks—a whole range of systems.

Q: To what extent have the tools you've been talking about helped to integrate test much earlier in the design process?
A: Whenever you have models of the various subsystems of your design, and then connect and simulate them virtually with Simulink, you have achieved much of that goal of introducing test earlier in the design cycle. Our new Simulink Design Verifier takes this process a couple steps further by enabling users to describe requirements for a design in the model itself.

In addition, Simulink Design Verifier will mathematically confirm that those requirements are met in the design. Another major benefit of Design Verifier is its ability to automatically generate test data that helps you to more comprehensively test the design. These powerful mathematical techniques, combined with simulation, bring testing to bear earlier in the development process, helping to achieve a more robust design.

Read the continuation of this interview.