Appendix D
Discussion About EE313

D.1 Introduction

Currently, eight instructors regularly teach EE313: Aggarwal, Araposthatis, Brown, Evans, Pearce, Sandberg, Smith, and Womack. These eight instructors have reached the following consensus about EE313:

1. EE313 has too many topics to be covered for a three-credit course;
2. M427K should not be the sole exposure to Laplace transforms that EE students receive;
3. it is possible to distribute the coverage of Laplace transforms and transfer functions among EE411, M427K, EE313, and EE438, which would enable a layered presentation of Laplace transforms; and
4. EE313 students lack a solid background in mathematics

This appendix summarizes the discussion about updating the list of topics to be more conducive to coverage in a single semester and to be more appropriate for training the students for later courses that depend on EE313. Section D.2 gives the list of topics for EE313 in the 1998-2000 catalog. Section D.3 suggests a shorter list of topics that would allow the topics to be treated in depth. Section D.4 discusses a proposed decoupling of EE313 and EE411, which was opposed 6-1 by the EE313 instructors (there was one abstention).

D.2 Topics for EE313 in the 1998-2000 Catalog

Signal representation; sampling and quantization; Laplace and z-transforms; transfer functions and frequency response; convolution; stability; Fourier series; Fourier transform; and applications Prerequisite: EE 411 [Circuit Theory] and Mathematics 427K [Advanced Calculus for Applications I] with a grade of at least C in each.

The following courses list EE313 as a pre-requisite:

• EE321 Electrical Engineering Laboratory I
• EE332K Numerical Techniques **
• EE345S Real-Time Digital Signal Processing Laboratory (formerly EE379K-17)
• EE351K Probability and Random Processes
• EE351M Digital Signal Processing
• EE362L Power Electronics **
• EE368 Electrical Power Transmission and Distribution **
• EE369 Power Systems Engineering **
• EE371M Communication Systems

D.3 Alternate list of topics for EE313

• EE438 Laplace transforms, transfer functions, frequency response
• EE351K continuous-time convolution
• EE362K Laplace transforms, transfer functions, frequency response, stability
• EE345S discrete-time convolution, z-transform, frequency response, AM/FM modulation
• EE351M discrete-time convolution, z-transforms, frequency response
• EE371M convolution, Fourier transforms, freq. response, AM/FM modulation

At Feb. 15th meeting, based largely on the above list, the attendees suggested the following changes for the list of topics for EE313:

Of the eight faculty who regularly teach EE313, two have objections to removing Fourier series and one agrees with the idea.

I would like to suggest that we add "systems" and "system properties" to the list of EE313 topics since I believe that we are teaching these topics anyway.

I would also like to suggest that M340L Matrices and Matrix Computation be a pre-requisite for EE313 because it would help students learn

1. state-space representation of systems (optional topic in EE313)
2. eigenfunctions for LTI systems (required topic in EE313)
3. matrix-oriented software such as Matlab (optional part of EE313)

Based on these suggestions, an alternate list of topics might be the following:

313. Linear Systems and Signals. Representation of signals and systems; system properties; sampling; Laplace and z-transforms; transfer functions and frequency response; convolution; stability; Fourier transform; AM/FM modulation; and applications. Prerequisite: EE 411, Mathematics 427K, and Mathematics 340L with a grade of at least C in each.

D.4 Discussion of Changes to EE313

Prof. #1 -- I found 313 to be a challenge: maybe the author(s) of the new 313 had limited experience in these courses. Nonetheless I approached the teaching of 313 as a challenge to the students and instructor. I have been satisfied with my results in 313: the pace is fast, but the students seem to like learning many topics even though rather shallow. I would think we should not tinker with 313 for several semesters until the track record becomes more apparent.

Concerning the when/where of Laplace transforms I find students in EE 362K want more coverage since "it has been several semesters since I had Laplace transforms". I guess the practical solution is tiered presentations and personal review.

Prof. John Pearce -- If 313 is too full we have 3 choices: (1) take out one or so of the topics (say, Z-Transforms); (2) make 313 into 413; or (3) create an additional course. I favor the second option of those two since Fourier Laplace and Z Transforms go well together. Alternately, if you want to put Fourier series representation into 411 that is not a bad idea, but some other things have to come out of it. Then where do they go? Dumping any of the topics in 411 is a bad idea as they all contribute significantly to following courses.

The third (not-popular-at-all) option is to create an additional course in the sequence. Say, one course in the minor sequence on continuous domain signal processing, as 313 could become (convolution, basis function representation, Fourier series, Fourier and Laplace transforms and circuits at multiple frequencies, audio, video etc.) and a second course in the major sequence (EE 32X on digital domain signal processing (sampling theory, digital & circular convolution, Z-transforms, Walsh basis functions, digital filters, digital audio & TV).

I feel very strongly that adequate time on these topics is crucial for our students as the concepts are central to EE practice. I fear that we have already jammed too much into the minor sequence ... and there are not superfluous topics there which can be combined or reduced in good conscience. I also would plead for some relief in EE 313, as I agree full well that the palate is heavily loaded for a 3 credit course. We had the present 313 topic list as a 4 credit course at Purdue and I found it a substantially heavy load at that time, in the now-ancient days of 1976.

Prof. #2 -- I agree EE 313 covers a lot of important material and there is not sufficient time to cover everything in any real depth. A big difficulty is that the students typically lack a solid background in mathematics. We could teach the math in EE 313, but this takes time. One solution is to in effect make EE 313 a two-semester course.

Concerning the alternate list of EE313 topics, the first item "Signal representation; sampling and quantization", isn't there too much emphasis on digital-signal processing oriented material? These things are better appreciated by students after they have understood the fundamentals of continuous-time systems (which are still very pertinent in engineering studies -- such as in the control systems area, in the areas of biology and medicine, in studies involving electromagnetic fields, in studies of transistor networks, etc.).

Concerning Laplace transforms, do we really want to relegate that important subject to mathematicians? There is certainly no harm in having mathematicians discuss them, but I think we (i.e., EE's) have a much better perspective concerning their utility and engineering limitations.

Prof. David Brown -- I think the basic problem is that we are attempting to cover the current list of EE 313 topics at too early a stage in the student's development. The student's have simply not had the time for fundamentals to "sink in." I include here fundamentals from the math courses and from first EE courses. I taught EE 313 for the first time last semester (I have taught EE 323 a number of times in the past). I was repeatedly surprised by the lack of mathematical sophistication and lack of circuit sophistication on the part of the students. I doubt if any re-sequencing of material will get around this problem if we serve up the material so early in the program. I expect that we will do so anyway since the prevailing mood seems to be in that direction.

Some specific thoughts: I do not think we can rely exclusively on the math courses to cover Laplace transforms. Our perspective is simply too different from that of the mathematicians. If we push Laplace transforms into EE 411, we will need to remove material from 411 which would then need to be covered elsewhere. In EE 313 last semester I followed the sequence of topics as presented in Lathi's book. This involved going back and forth between continuous systems and discrete systems. There are some advantages to doing this, but I think it's a rough ride for the students. This semester, I am sequencing things differently which means skipping around in Lathi's book (which is also objectionable, at least to some students). I am spending several weeks up front on Chapters 1 and 2 supplemented with my own approach to state variables and basic frequency response and filtering concepts. The frequency response and filtering concepts follow easily from the way I do EE 411. I then plan to do Fourier series, Fourier transforms, and Laplace transforms in that order. I like to present Fourier transforms by extending Fourier series; Laplace transforms by extending Fourier transforms. Then we will pick up discrete systems and z transforms. I hope this works better than what I did last semester, but it's too early to tell.

I hate to see us "decouple" 411 from 313. I know that 313 is supposed to be a signals and systems course and not a circuits course. However, circuit examples presented in EE 313 can make the material a lot more interesting and relevant.

Prof. Aristotle Araposthatis -- I agree that current syllabus of EE313 lists too many topics. A possible solution is to revise 302 and 411 so as to cover Laplace transforms (or, to be exact, elementary symbolic calculus with Laplace transforms) in EE411. I am a routine instructor of EE411 and in my experience this would be feasible, provided some of the basics of RLC networks are covered in EE302.

Concerning EE313 preceding EE411 in the curriculum and being transformed essentially to a signal-processing course I have some rather strong objections, summarized below:

• To attempt to cover topics in 313, without the students even having a clue how to realize a simple filter with an RLC network, would run into serious problems. To try to teach Fourier theory, however elementary, to students that have not seen the derivation of the steady state sinusoidal response for a second order RLC, I think is unnatural. Note, that before the most recent revision of the curriculum EE411 was a two semester sequence, EE411A and EE411B. A good part of the material of EE411B is now covered in EE313. I am aware that Ed Lee at Berkeley has organized an elementary Signal Processing course, heavily using Matlab. I don't know if the inspiration for the suggested changes to EE313 comes from this direction. Do we have a syllabus for this course handy, and do we know if it precedes Circuits in the Berkeley curriculum?
• On the issue of EE313 becoming an introductory digital signal processing course, my main objection is that the courses that we should include in the basic sequence should cover "generic" material and provide the students with a solid understanding of fundamental principles and techniques. Drifting towards a signal processing course, would deviate from this objective. I think that a signal-processing course should be not be offered until later, perhaps as a block elective.
• Concerning relying on the Math Department for teaching our students Laplace transforms, I think we shouldn't. In my experience, students that have just finished a course in Linear Algebra in the Math Dept. seem to lack knowledge of such basic concepts and techniques as "determinant", "matrix by vector multiplication" or "Cramer's rule". A possible explanation is that the students find it very hard to bridge the gap between Linear Algebra taught in an abstract setting and the elementary calculations needed in Engineering Courses. It is better if we work it out to include Laplace Transforms in EE411. I think it can be done.

Prof. #3 -- I agree with the comments of Prof. Araposthatis.

Last updated 11/09/00. Mail comments about this page to bevans@ece.utexas.edu.