active noise control: is it good for anything? · dept. of physics & astronomy brigham young...
TRANSCRIPT
May 22-23, 2011
Active Noise Control: Is it Good for Anything?
Scott D. Sommerfeldt Acoustics Research Group
Dept. of Physics & Astronomy
Brigham Young University
April 2, 2012
May 22-23, 2011
Physics and Astronomy Department Size
32 Faculty Members
325 Undergraduates
35 Graduate Students
Acoustics AMO Astronomy/Astrophysics Condensed Matter Plasma Theoretical Physics
May 22-23, 2011
Acoustics Research Group
• Interdisciplinary Group – currently primarily Physics and ME
• Physics: 4 full-time faculty (Sommerfeldt, Leishman, Gee, Thomas), one part-time faculty (Neilsen)
• ME: 2 faculty (Blotter, Thomson)
• Typically ~15 grads and ~ 25 undergrads involved from Physics, ME, few from EE
• ARG meets weekly to discuss acoustics topics – typical attendance is about 30
• External funding for group ~ $400k/year
• Strong student chapter of Acoustical Society of America
• Program recognized widely nationally
May 22-23, 2011
Outline
• History and background
• Principles of active control
• Local vs global control
• Global control for free-field radiation
• Global control in an enclosed field
• Conclusions
May 22-23, 2011
Brief History
• Active control began in 1930’s, with work of Paul Lueg
• Major drawback:
- acoustic feedback
- only worked at certain frequencies
- drifting electronics
A
λ/2
May 22-23, 2011
History (cont.)
• Brief revival in 1950’s – Harry Olson
• Development of digital signal processing led to explosion of research beginning around 1980.
• Number of papers increased nearly exponentially for couple of decades – appears to be leveling off some now.
May 22-23, 2011
Active Control Principles
• Most acoustics applications involve linear systems – principle of superposition holds.
• How well must the fields match?
• Phase errors: Magnitude errors: For -20 dB, θ = 5.7° For -20 dB, amp. ratio = 1.0 dB -30 dB, θ = 1.8° For -30 dB, amp. ratio = 0.3 dB
May 22-23, 2011
Mechanisms of Active Control
I. Destructive Wave Interference
• Interference between waves leads to "zone of silence."
• Size of zone of silence - 0.1 λ.
• Global energy in the field increases. May not be a problem if the increase in energy is localized to areas of little concern.
May 22-23, 2011
Mechanisms of Active Control (cont.)
II. Mutual Coupling • Closely spaced sources modify each other's
radiation impedance. • Possible to achieve global attenuation of sound or
vibration field. (Ex. dipole vs. monopole) • Power radiated by two closely spaced sources
proportional to 1 - sin(kd)/(kd). Thus, sources must be closely spaced.
• For enclosed fields, sources may be remote from each other, but still coupled through the modes of the enclosure.
May 22-23, 2011
Adaptive Minimization
• Schematic of control system
• Error given by: 1
0
)()()(N
n
n ntxwtdte
May 22-23, 2011
Adaptive Minimization (cont.)
• Update coefficients using negative gradient
• Result:
)()()()1( tentxtwtw nn
May 22-23, 2011
Global ANC
• Suppression vs. interference
Mutual coupling of sources affects radiation resistance, power radiated
• Multi-channel systems often required for three-dimensional sound fields
• Global control requires “small” separation between noise and control sources
• Choose error sensor locations and types which result in global attenuation
May 22-23, 2011
Difficulty of ANC Problems
13
May 22-23, 2011
Cooling Fan Noise
• Axial fans used to cool computers, projectors, etc.
• Spectrum dominated by tonal noise (BPF and harmonics)
• Tonal noise result of unsteady loading on the fan blades
May 22-23, 2011
Previous Research and Objectives
• Previous: Single channel efforts
Error microphone usually in far field
Significant global attenuation for BPF (~12 dB), less for harmonics
• Objectives: Multi-channel ANC to achieve global tonal
attenuation
Error microphones in “practical” near field locations
May 22-23, 2011
Theoretical Analysis
• Determine:
“good” near field error microphone locations for given control source configuration
appropriate number of control sources
• Model fan/control sources as ideal point sources
• Calculate control source strengths that minimize radiated power
• Plot 2-D pressure and find potential microphone locations
May 22-23, 2011
Theoretical Analysis Results
May 22-23, 2011
Theoretical Analysis Results
May 22-23, 2011
Experimental Apparatus
• 80 mm (3.25”) seven-bladed fan Mounted on face of “CPU-like” enclosure
1” wide rectangular obstruction to create unsteady loading on fan inlet
BPF maintained at 370 Hz
• Four 1 1/8” loudspeakers mounted around fan
• Error microphones located on surface of enclosure
• Nonacoustic reference sensor Infrared emitter/detector
Voltage waveform with BPF and harmonics
• Multi-channel filtered-x LMS algorithm
May 22-23, 2011
Experimental Apparatus
May 22-23, 2011
Reference Signal Power Spectrum
May 22-23, 2011
ANC at Error Microphone
May 22-23, 2011
2 x 2 ANC—Ideal
7.1 dB 13.5 dB
5.6 dB 5.4 dB
Mean-Square
Pressure Reduction
May 22-23, 2011
4 x 4 ANC—Ideal
8.4 dB
15.5 dB 14.3 dB
18.5 dB
May 22-23, 2011
Cooling Fan Control System
• Previous System
– 80 mm fan
– 32 mm loudspeakers
– 370 Hz BPF
• Modifications
– 60 mm fan
– 20 mm loudspeakers
– 600 Hz BPF
– Increase fan speed
May 22-23, 2011
60 mm Fan Results
May 22-23, 2011
Global Comparison
80 mm 60 mm Ideal
kd MPR kd MPR kd MPR
BPF 0.4 10.1 dB 0.5 13.6 dB 0.5 ~30 dB
2 x BPF 0.8 16.1 dB 1.0 17.6 dB 1.0 22 dB
3 x BPF 1.2 12.8 dB 1.5 10.5 dB 1.5 14 dB
Mean-square pressure reduction (MPR)
May 22-23, 2011
Optimizing the Source Configuration
• Symmetric source configuration shown was assumed to be optimal
• Configuration can be optimized through use of a genetic algorithm
• Four control sources assumed
May 22-23, 2011
Optimizing the Source Configuration
May 22-23, 2011
Symmetric vs. Linear Configuration
May 22-23, 2011
Fan-Noise Findings
• Multi-channel ANC for cooling fan successfully demonstrated
• Point source modeling yields consistently good near field locations for error microphones
• Fan is complex noise source at higher harmonics—need multiple control channels for global reductions
• With better loudspeakers, can achieve more attenuation for BPF
80 mm fan 60 mm fan
May 22-23, 2011
Active Control of Energy Density in a Vehicle Cabin
May 22-23, 2011 The operator’s head does not necessarily stay in a fixed location
during typical operation.
May 22-23, 2011
Energy Density vs Pressure2
• Squared Pressure (SP) – More spatial variation
– Control can yield low pressure but potentially high velocity
– Control tends to be more local
• Energy Density (ED) – Less spatial variation
– Control requires both pressure and velocity to be minimized
– Control tends to be more global
May 22-23, 2011
• Kinetic Energy
• Potential Energy
• Total Energy
• Instantaneous Energy Density
Acoustic Energy Density
May 22-23, 2011
Spatial Dependence for a Rectangular Enclosure
May 22-23, 2011
SP Nodal Structure
May 22-23, 2011
ED Nodal Structure
May 22-23, 2011
2D Sensor
•2D Sensor Effective When Placed Near
Wall or Ceiling
•Computationally More Efficient
May 22-23, 2011
The Enclosure
•Potential
Application: Cab
Noise
•Enclosure Fairly
Representative of
Small Equipment
Cab
May 22-23, 2011
Sensor and Mic Placement
•ED Control Has Less Dependence on
Sensor Location
May 22-23, 2011
Noise Source and Subwoofer
May 22-23, 2011
110 Hz Tone @ Sensor
May 22-23, 2011
110 Hz @ Ear Location
May 22-23, 2011
Engine Tone Attenuation (dB)
46 Hz 90 Hz 110 Hz 120 Hz
12-Mic Avg 33.47 16.55 23.97 20.62
Right Ear 34.95 22.24 30.48 30.60
Left Ear 35.02 19.41 23.68 26.72
Sensor 34.49 22.58 32.92 32.05
May 22-23, 2011
Sound Level Reduction (dBA)
46 Hz 90 Hz 110 Hz 120 Hz
12-Mic Avg 4.11 0.39 6.12 2.77
Right Ear 3.84 0.60 8.71 4.75
Left Ear 4.27 0.57 8.70 4.69
Sensor 2.6 0.36 7.68 4.48
May 22-23, 2011
Cab Noise Findings
• Engine tones can be attenuated drastically
• Attenuation is achieved throughout cabin
• Significant reduction in total sound level is possible, even with A weighting
• Much can still be done with respect to performance optimization
50 Hz
May 22-23, 2011
Examples
• Heavy Equipment (Caterpillar)
– Wheel Loader Cab
– Focused on tonal noise at engine firing frequency
48
May 22-23, 2011
49
Hardware Configuration
• Sub-woofer used for low frequencies
• Smaller drivers (10 cm) used for higher frequencies
• ED sensor mounted in convenient location
May 22-23, 2011
50
Cab Response – Engine Sweep
• Results measured at nominal operator head location
Relative Lp
Relative Lp
50
25
0
50
25
0
May 22-23, 2011
51
Control Results – Engine Sweep
Relative Lp
50
25
0
Relative Lp
50
25
0 0
May 22-23, 2011
52
Engine Firing Freq – Engine Sweep
972H prototype - firing frequency Lp at left ear
40
50
60
70
80
90
100
950
1150
1350
1550
1750
1950
2150
rpm
Lp
(d
B)
ANC OFF
ANC ON
10 d
B/d
iv
May 22-23, 2011
53
Cab Results – Global Response
-3.4 -2.2 0.1
-6.3 -6.4 -5.0
-8.1 -7.5 -5.2
operator
head
upper plane
front
0.2 0.9 -1.1
-2.5 -3.9 -1.3
-5.7 -5.0 -5.2
lower plane
May 22-23, 2011
Conclusions
• Active noise control is a viable solution for certain applications.
• Need to be careful in applying technology in order to achieve desired results.
• Good understanding of physics of application necessary for proper implementation.
• Technology starting to find its way into real applications – cost is still an issue but it has come way down.
May 22-23, 2011
Current Applications
• Honda 2005 Odyssey and Accord Hybrid
• Lexus
• Accura
• Saab commuter jet
• Active headsets
• BMW working on implementing an ANC system
May 22-23, 2011
Special Thanks To . . .
• Kent Gee
• Ben Faber
• Brian Monson
• Connor Duke
• Ben Shafer
• Jared Thomas
• Stephan Lovstedt
• The BYU Acoustics Research Group