Session 1320

Design of an Electronic Muffler - A DSP Based Capstone Design Project

George Piper, John Watkins, Carl Wick, Svetlana Avramov-Zamurovic
United States Naval Academy

Abstract
Active control of noise has been an emerging technology for the past two decades. Active noise
control (ANC) is an attractive means to achieve large amounts of noise reduction in a small
package, particularly at low frequencies, where passive noise control may be impractical. While
the concept of ANC has long been established, the technological means for implementing ANC
have only recently become available. With the advent of high-speed digital signal processors
(DSPs) and modern signal processing methods, ANC is now becoming a reality.

This paper presents an interesting undergraduate design project involving ANC. In their
capstone engineering design course students designed and built a prototype ANC system that
successfully demonstrated the concepts of an electronic muffler. The heart of the project
centered on the popular Analog Devices’ SHARC DSP evaluation module (EVM). In this paper
we will discuss ANC concepts and the electronic muffler, the students design project, and the
pedagogical outcome.

1.0 Introduction
The Systems Engineering curriculum at the U.S. Naval Academy is a four-year, undergraduate,
ABET accredited, program, where students specialize in the interaction between mechanical,
electrical, and computer systems. The curriculum focuses mainly on linear systems theory,
feedback control, and mechatronics. An integral component of the Systems Engineering
curriculum is its capstone design course. This course provides students with a comprehensive
design experience bridging the gap between textbook problems and real life applications.
Students work in teams to complete a semester-long design project. Each team must propose a
project, prepare schedules and reports of their activities, and design, build, and test a prototype.
One of the many challenges facing students in this course is the selection of a project that is
meaningful and representative of their engineering discipline.

For nearly a decade automotive manufacturers have been experimenting with electronic mufflers
for automobile exhaust systems [4] and their realization is near production. Some current
electronic muffler concepts are based on active noise control (ANC) that uses destructive
interference of sound waves to reduce unwanted noise. An additional advantage of an electronic
muffler is the reduction of backpressure associated with conventional mufflers. Reduction of
backpressure provides improved fuel economy in most gasoline engines.

A relevant on-going project in the Systems Engineering Dept has been the design of an
interference-based electronic muffler. The Systems Engineering Department has had several
teams of students over the past three years working on electronic muffler designs. Students find
the design of an electronic muffler project intriguing because draws upon previous course work
Page 6.338.1
in systems engineering and challenges them with advanced topics in active noise control and in
digital signal processing.

The design and construction of a fully functional electronic muffler is impractical for a one-term
design project. However, students do not need to solve every aspect of a complex, real-world
project to make it a meaningful design experience. Rather, students can identify one or two key
design issues, whose resolution would contribute to the overall objective of the project [1]. The
key issue in the design of an electronic muffler is the design of the active noise controller.
Therefore the focus of the electronic muffler project is for students to design a prototype ANC
system that simulates the behavior of an electric muffler.

In the sections that follow, we will discuss the ANC and DSP aspects of this project, the
student’s electronic muffler design, and the pedagogical outcome of this project.

2.0 The Electronic Muffler Concept and Active Noise Control

An electronic muffler is a practical example of ANC technology. Noise radiating from the outlet
of an automobile exhaust system is reduced by transmitting "anti-noise" from a loud speaker
located near the end of the exhaust pipe. The acoustic output of the loud speaker is controlled so
that it destructively interferes with the automotive noise. The concept of an electronic muffler is
shown in Figure 1.

Active Noise
Controller
Tach

Speaker Microphone

Automobile
Engine Exhaust Pipe

Figure 1. Electronic Muffler Concept

Much of the noise generated by an internal combustion engine is periodic and its frequency
components vary as a function of engine rotation speed. An active noise controller uses a
tachometer to synthesize a reference signal that is coherent with engine rotation speed. The
reference signal is then processed with an adaptive digital filter to produce a signal that is 180°
out-of-phase with engine noise. This signal drives a loud speaker to generate anti-noise. An
error microphone located at the outlet of the exhaust system measures residue engine noise. This
error microphone signal is used to automatically adjust the digital filter’s coefficients such that
the residue noise is minimized [3][4]. The most common form of adaptive algorithm used in
ANC is the least mean squares (LMS) algorithm developed by Widrow [5].

While the concept of ANC has long been established, the technological means for implementing
ANC has only recently become available. With the advent of high speed digital signal
processors (DSPs) and modern signal processing methods, ANC is now becoming a reality.
Page 6.338.2
3.0 A DSP Based Capstone Design Project
As mentioned earlier, the objective of the electronic muffler project is to have students design a
prototype ANC system that simulates an electronic muffler. The project is ideal for a team of
two Systems Engineering students. Students who have worked on this project in the past were
familiar with linear systems theory, digital control systems, basic electronics, and elementary
physics of waves. Prior to the project, these students had no exposure to active noise control or
to adaptive signal processing. These new concepts were introduced to the students by the
project’s faculty advisor. Also, as part of the project, the students did extensive research on
these topics.

Over the years, there have been several iterations of the electronic muffler design in this project.
They have ranged from trombone-like devices to simple speaker and pipe arrangements.
However, the most recent and most successful design is depicted in Figure 2.

Acoustic Duct
Noise Noise Error
Generator Source Anti-Noise Microphone
Source
Reference
Signal

D/A A/D

FIR
x Filter y

Adaptive D/A e
Update
DSP Controller
Figure 2. Project System Diagram

For their design, the students proposed to build a prototype ANC system that would demonstrate
narrow band noise attenuation in the 200 to 800 Hz frequency range. The ANC system that
materialized consisted of two speaker enclosures connected by a length of 1 1 2 ” diameter PVC
pipe, a microphone, a digital signal processor, signal conditioning circuitry, and power amplifiers
to drive the speakers. A photograph of the student’s project is shown in Figure 3. One speaker
acted as the noise source simulating the noise generated by an internal combustion engine. The
other speaker acted as the anti-noise source. The PVC pipe represented the automobile exhaust
pipe. A signal generator was used to drive the noise source speaker. An Analog Devices
SHARC DSP was programmed as an active noise controller that generated the anti-noise signal. Page 6.338.3
 Figure 3. Electronic Muffler Project

The heart of the project centered on the implementation of an adaptive digital filter for the active
noise controller. The adaptive filter used in this project consisted of two distinct parts as shown
in Figure 2: a digital filter to modify the amplitude and phase of the reference signal x(k), and an
adaptive algorithm for adjusting the filter’s coefficients. The digital filter was implemented as a
finite impulse response (FIR) filter. Because a FIR filter has only zeros and no poles itis always
stable. The output of the digital filter y (k) at time k is computed as

y ( k ) = w T ( k ) x( k )

where x is the input vector at time k

x(k ) = [ x(k ) x(k − 1) K x(k − N + 1)]

T

and w is the time-varying coefficient vector at time k

w (k ) = [ w0 (k ) w1 (k ) K wN −1 (k )]
T

A least-mean-squares (LMS) adaptive algorithm [5] is used to adjust the filter coefficients such
that the noise detected by the error microphone e(k) is minimized. The filter coefficients are
updated every control cycle according to

w ( k + 1) = w ( k ) + µ ⋅ w ( k ) ⋅ e ( k )
Page 6.338.4
where µ is the step size. By adjusting µ , the algorithm performance characteristics such as
stability and convergence time of the coefficients can be changed.

The students implemented their filter on an Analog Devices ADSP-21061 SHARC EZ-KIT
evaluation board. The SHARC evaluation board was recommended for the project by the faculty
because it is an affordable and effective tool for developing DSP applications. Students used the
(locally generated) SHARCByte user interface [6] to write a C program to implement the
adaptive filter algorithm on the SHARC processor.

The students examined the ANC system’s performance for tonal noise reduction over the
frequency range of interest (200 - 800 Hz). They measured noise attenuation by comparing the
error microphone output when the active noise controller was operating against when the active
noise controller was not operating. They found that the active noise controller performed well
over certain frequency bands. Outside these frequency bands the adaptive digital filter would not
converge and the system would become unstable. The students concluded that this anomaly was
caused by the adaptive filter’s inability to make large changes in phase shift. They
recommended adding predetermined time delays for different frequency bands to introduce the
needed phase shift. The student’s recommended fix is similar to filtered-x LMS algorithms
found in the literature [3].

4.0 Conclusions
In this paper a DSP based project for a capstone design course was discussed. The objective of
the project was to design a prototype ANC system that demonstrates the concept of an electronic
muffler. Experience has shown that this project provides the students with a meaningful
comprehensive design experience that bridges the gap between textbook problems and real life
applications. The project not only draws on students previous course work in systems
engineering, but it also introduces students to advanced topics in digital signal processing and
active noise control.

5.0 References
1) Marin, J.A., Armstrong, J.E., Kays, J.L., "Elements of an Optimal Capstone Design
Experience", Journal of Engineering Education, January 1999, pp. 19-22.

2) Wankat, P.C., and Oreovicz, F.S., Teaching Engineering, McGraw Hill, New York, NY, 1993.

3) Nelson, P.A., & Elliott, S.J., Active Control of Sound, Academic Press, San Diego, CA, 1994.

4) Kuo, S.M., & Morgan, D.R., Active Noise Control Systems / Algorithms and DSP
Implementations, Wiley Interscience, New York, NY, 1996.

5) Widrow, B., and Stearns, S.D., Adaptive Signal Processing, Prentice Hall, Englewood Cliffs,
NJ, 1985.

6) Wick, C.E. and G.E. Piper, “Using the ADSP-21061 SHARC EZ-KIT in Undergraduate DSP
Oriented Courses”, Proceedings of FIE98, Session F2D, CD-ROM ISBN 07803-4765-X.

6.0 Biographical Information

GEORGE E. PIPER is an Associate Professor of Systems Engineering at the United States Naval
Page 6.338.5

Academy. He joined the Naval Academy faculty in January 1994. He holds a B.S., M.S. and a
Ph.D. in mechanical engineering from Drexel University. Prior to joining the Naval Academy
faculty, Dr. Piper was a senior member of the technical staff at Martin Marietta’s Astro Space
Division. He is a licensed Professional Engineer. Dr. Piper's current interests include active
noise control, space vehicle dynamics and control, and engineering design.

JOHN M. WATKINS is an Associate Professor in the Systems Engineering Department at the

United States Naval Academy. He received his B.S. degree in electrical engineering from the
University of Nebraska-Lincoln in 1989 and his M.S. and Ph.D. degrees in electrical engineering
from the Ohio State University in 1991 and 1995, respectively. His research interests include
feedback control systems, system identification, fault detection and isolation, magnetic
levitation, and active noise control.

CARL E. WICK received the BS degree from the U.S. Naval Academy in 1970, the MS degree
from the Naval Postgraduate School in 1976 and the Doctor of Science degree from the George
Washington University in 1993. A retired career Naval Officer and aviator, he has been with the
Weapons and Systems Engineering Department of the U.S. Naval Academy since 1990. His
current research interests include embedded computer systems and simulators, digital signal
processing, and image processing systems. He is a member of SPIE, Sigma Xi, and is an
associate editor for ASEE Coed.

SVETLANA AVRAMOV-ZAMUROVIC received the Ph.D. in Electrical Engineering from the

University of Maryland in 1994. At present she is an Associate Professor with the Weapons and
System Engineering Department at the United States Naval Academy. From 1990 to 1994 Dr.
Avramov-Zamurovic was involved in developing a measurement bridge for voltage ratio
calibration for the NASA space experiment Zeno. She was a guest researcher at the National
Institute of Standards and Technology (NIST) from 1990 to 1994. During the summer of 1995
she was involved in development of capacitance ratio bridges to support the Single Electron
Tunneling Experiment at NIST. Her research interests include precision measurements of
electrical units, in particular the development of bridges to measure impedance and voltage
ratios.

Page 6.338.6