Let modeling solve test challenges
An exclusive interview with a technical leader
By Larry Maloney, Contributing Editor -- Test & Measurement World, 9/1/2007
|
Q: Where does The MathWorks fit in the world of engineering?
A: The company was founded in 1984 around our Matlab software, so technical computing is a strong suit: bringing in data and analyzing it and developing and deploying algorithms and technical applications. Over the last 15 years, another key area has been model-based design, which is changing the way engineers develop, test, and verify components and systems.
Jim Tung provides more specifics on his company's products and test applications, as well as its development strategies, in the continuation of this interview.
Q: How are engineers using Matlab for test and measurement?
A: One of the strengths of Matlab for test is its ability to analyze and model data gathered from hardware and software. Increasingly, vendors of bench and ATE equipment are incorporating Matlab in their products as instrumentation and test systems become more PC-based. This enables engineers to use Matlab on the instrument to construct custom filters or modulation waveforms or to add analysis routines.
We’ve also developed several modules for Matlab targeted for test and measurement. These include the Data Acquisition Toolbox that connects Matlab to data-acquisition boards on the PC and the Instrument Control Toolbox that links Matlab to test hardware through GPIB, RS-232, and other interfaces. Also very important is our System Test tool. It helps engineers working in Matlab or Simulink, our block-diagram modeling environment, to start building test routines early in the development process.
Q: Why the big focus on modeling?
A: Modeling tools, such as our Simulink family, are becoming a core part of the design and verification process. System design increasingly involves creating models of physical components that affect system behavior—the mechanics, electronics, controls—as well as the environment. In an automotive suspension system, for example, that environment includes the road surface, weather conditions, and the driver. Finally, the model includes the algorithms or logic used in the system.
You can also automatically generate embedded software from the algorithm models for your system, and with our new Simulink HDL Coder tool, you can even generate Verilog or VHDL code for digital electronics. This saves a huge amount of time and money, versus creating multiple physical prototypes and manually writing the code. Meanwhile, the model allows continuous validation and verification of your design, and it can be leveraged to design an ATE or hardware-in-the-loop setup that includes the behavior of the system in operation.
Q: How does your Simulink family make system modeling easier?
A: Simulink enables engineers to accurately model, simulate, and implement the behavior of a variety of continuous or discrete time systems. These systems include controls, signal processing, image processing, communications, and physical objects. Our users have created whole libraries of Simulink models that other engineers can use to accelerate the development process.
Q: What impact is modeling having on the test-engineering function?
A: Traditionally, the test engineer tended to be at the end of the line, designing tests based solely on product or system specs. Now, test engineers are increasingly using design models as the basis for designing their tests. We also see more design engineers doing some test work early in the process, such as running Monte Carlo simulations or hardware-in-the-loop tests. In fact, The MathWorks is trying to accelerate this trend with the a new product called Simulink Design Verifier.
Read the continuation of this interview.


























