At the Georgia Institute of Technology, or Georgia Tech, professors within the College of Engineering have long recognized the value of teaching computer science concepts to their students. However, developing a course that enabled students to learn and apply computer science fundamentals to solve engineering problems has not always been easy.

"We have tried various ways to develop a computing course, but we were never satisfied with the results," says Dr. James Craig, professor of aerospace engineering at Georgia Tech. "We realized that we were teaching tools, but were missing the computer science concepts."

Recently, the College of Computing worked with the College of Engineering to develop a practical introduction to computer science for engineers using MATLAB. Today, each semester, more than 1,000 students take the course, CS1371 - Computing for Engineers.

MATLAB enables students to learn the computer science concepts in a way that is directly applicable to engineering. MATLAB is an excellent first language for engineers and an ideal environment for engineering computation," says Craig. "In the upper-level engineering classes, we are very impressed with the capabilities of the students now. The course has been a huge success."

Challenge

Instructors and students struggled with earlier versions of the course, which were based on languages such as Pseudocode and Scheme. "Our approaches were missing something," notes Craig. "If we needed to do matrix multiplication or solve simultaneous equations, we were stuck because we had to write our own low-level functions for those tasks."

The college also wanted the course to engage the engineering students and enable them to develop skills that they could leverage throughout their undergraduate studies and beyond.

"With MATLAB, we are combining computer science theory and concepts with problem solving in engineering. MATLAB is the one language that we want our students to use-the one that we all use in our classrooms."

Solution

Georgia Tech adopted MATLAB as the foundation for CS1371 - Computing for Engineers. Taught by the College of Computing, this course is required for all Georgia Tech engineering students and is now a pre-requisite for many advanced level courses.

Instructors teach the course with MATLAB open, writing code to illustrate new concepts as they are discussed.

Concepts in Computer Science Implemented in MATLAB, written by David Smith, a primary instructor for the course, guides the coursework. The course is delivered in three segments: basic procedural programming, writing applications, and dynamic data structures.

The first three weeks are devoted to writing functions and scripts and operating on vectors, conditionals, loops, and iteration. The first segment concludes with structures, arrays, character strings, and an introduction to recursion.

In the next segment, the students begin working with plots and images. "After week six, the applications become more relevant to the students' interests," notes Smith. "The students learn how to do two- and three-dimensional parametric plotting with MATLAB."

The course also dedicates a week to numerical methods, applying MATLAB functions before concluding the second segment with a discussion of sorting.

In the third segment, devoted to dynamic data structures, the students implement linked lists, binary trees, N-ary trees, and graphs in MATLAB.

"Students really identify with visual projects, so we work with images," notes Smith. "For example, I show how to find the gray sky in a picture using MATLAB, and replace it with a blue sky without destroying the rest of the picture."

After completing the course, Georgia Tech engineering students continue using MathWorks tools throughout their undergraduate studies. Instructors in virtually every engineering discipline leverage tools such as Simulink, the Control System Toolbox, and Real-Time Workshop.

Results

Products Used

• Control System Toolbox
• MATLAB®
• Real-Time Workshop®
• Simulink®

From the Eiffel Tower to the Sydney Opera House, high-intensity discharge (HID) lamps illuminate some of the most recognizable landmarks in the world. Such large areas typically require more light than can be provided by conventional incandescent or fluorescent lamps. HID lamps are optimal for lighting vast outdoor spaces because they deliver high light output. For indoor applications, HID lamps are primarily used for accent and decorative lighting.

A division of Royal Philips Electronics, Philips Lighting, together with partners in the European Lamp Manufacturers Association in the Preparation of Standards (ELMAPS), have taken a leadership position in delivering HID lighting solutions and ensuring safety and performance through industry standards for electronic ballasts. Philips Lighting engineers use MATLAB to enable other lighting component manufacturers to verify that a key performance critical aspect of electronic HID ballasts, ripple power, is within acceptable limits.

"We wanted to create a widely available tool to help companies ensure compliance with this standard," says Don Couwenberg, global HID consultant at Philips Lighting. "Working from a sophisticated mathematical description of the problem, we used MATLAB to build a standalone tool that everyone can use easily."

Challenge

HID lamps incorporate a ballast, circuitry that delivers a high-voltage pulse during startup and limits lamp current once it is in steady state. The ballast is a switch-mode power supply that transforms power from the main power source and produces the voltages and nominal current needed by the lamp. As a byproduct, the ballast typically also produces a ripple current at frequencies higher than 10 kHz. When the associated power with this ripple current exceeds 1.5 percent of the nominal power of the lamp, the lamp can become unstable and potentially fail.

Philips Lighting, Osram, GE, and Sylvania agreed on a method of computing the ripple current in a lamp based on measured lamp voltages and current over time. Warren Moskowitz of Osram proposed the method, which was presented in 2004 at the Lighting Conference LS10 in Toulouse, France. While acquiring the data required for this computation is fairly straightforward, the process for determining the ripple current is complex. Philips Lighting needed a tool to calculate ripple power from measured data. They also needed the tool to be easily distributed to and employed by ballast manufacturers.

"We want to promote this standard so that our next generation of products can incorporate components from other vendors. We need specifications so that there are no incompatibilities, the lamps perform optimally and safely, and our customers get the full benefits of standardization," says Couwenberg.

"At Philips Lighting, MATLAB and Simulink are key tools for mathematical analysis and computation. For HID lamps, MATLAB plays an important role in measuring ripple power and ensuring that performance standards are met in the industry."

Solution

Using MATLAB and the MATLAB Compiler, Philips Lighting created a standalone application for calculating ripple current and ensuring compliance with industry standards.

Couwenberg started by engaging MathWorks Consulting to develop a first proof-of-concept version of the application using MATLAB. Working from this version, Philips Lighting engineers continued using MATLAB to develop a full-featured version.

The engineers used MATLAB to implement an algorithm that analyzes a series of lamp voltages and lamp current measurements taken at periodic intervals to determine the harmonics of the ripple current.

The group used the Signal Processing Toolbox to perform the fast Fourier transform and perform sensitivity analysis by evaluating various windows, such as Blackman and Hamming windows.

They also used MATLAB development tools to create a graphical user interface that displays the harmonics as well as a test pass/fail indicator that verifies ballasts are within acceptable standards when the computed ripple power is less than 1.5 percent of the nominal power.

Engineers then used the MATLAB Compiler to create a standalone executable version of the tool that can be run outside MATLAB, saving them time in rewriting the application.

Philips Lighting is in the final stages of prerelease testing of the tool. They are also using Simulink and additional MathWorks tools to model sophisticated control loops for the switch mode power supply of lamps to enhance color performance and light quality.

Results

Products Used

• MATLAB®
• MATLAB® Compiler
• Signal Processing Toolbox
• Simulink®