Presentation on theme: "ACTIVE CONTROL OF SOUND Professor Mike Brennan Institute of Sound and Vibration Research University of Southampton, UK."— Presentation transcript:
1 ACTIVE CONTROL OF SOUND Professor Mike Brennan Institute of Sound and Vibration Research University of Southampton, UK
2 Active control of sound Active control of sound in ductsSingle secondary sourceTwo secondary sourcesWhere does the power go?Control of harmonic disturbancesControl of random disturbancesSingle channel feedforward controlConstraint of CausalityActive control of sound in enclosuresCarsAircraftActive head setsVibroacoustic control
3 Passive Control of Sound Sound sourceObserverPassive control relies on barriers, absorption and damping.It works well when the acoustic wavelength is short compared with typical dimensions Higher frequency solution.
4 Active Control of Sound Sound sourceObserverAcoustic or structural actuators are driven to cancel waves:It works well when the acoustic wavelength is long compared with typical dimensions Lower frequency solution.
5 Patent for Active Control of sound by Paul Lueg 1936 Active Control of Duct-Borne SoundPatent for Active Control of sound by Paul Lueg 1936
6 Loudspeaker source in a duct If the frequency of interest is such that the acoustic wavelength is greaterthan twice the dust cross-section then it can be modelled as a pair of massless pistons forced to oscillate apart with a fluctuating volume velocity q(t) between them.
7 Loudspeaker source in a duct For x > 0 the complex pressure and particle velocity fluctuations can be written as:For x < 0where U+ is the velocity of the right-hand piston and U- is the velocity of the left-hand piston.
9 The plane monopole source We define the source strength asSo
10 Cancellation of downstream radiation using a single secondary source Choose a secondary source strength to set pressure field downstream of secondary source to zeroPrimary sourceSecondary sourceThe fields due to the primary and secondary source areUse the principle of superposition to calculate the net sound field
11 Cancellation of downstream radiation Primary sourceSecondary sourceThis requirement iswhich leads tothat is the secondary source is a delayed inverted form of the primary source.
12 The net sound field in the duct The field between the primary and secondary sources is give byUpstream of the primary source it is given byDownstream of the secondary source it is given by
13 The net sound field in the duct -110.51.522.5Note that when L=nλ/2 the pressure upstream of the primary source =0
14 Time domain interpretation Primary sourceSecondary source
15 Cancellation of downstream radiation using a pair of sources Primary sourceSecondary sourcesDownstream of the second secondary source the net pressure field can be set to zero by settingWith two sources it is possible to ensure zero radiation upstream of theSecondary source pair by setting
16 The net sound field in the duct The field upstream of the secondary sources is given byBetween the secondary sources it is given byDownstream of the secondary sources it is given by
18 Time domain interpretation The secondary sources are given byTo enable interpretation in the time domain let us choose a primary source strength whose Fourier transform is some function i.e.,In the time domain this assumesIt then follows thator in the time domainSo
19 Time domain interpretation Primary sourceSecondary sources
20 Sound absorption by real sources Electrical power suppliedElectricalimpedanceMechanicalimpedanceAcousticalimpedanceThe acoustical power can be negative; in such cases less electricalpower will be required to sustain a given piston velocity u
21 The influence of reflections from the primary source Absorbing surface havinga complex reflectioncoefficient RPrimary sourceSecondary sourceTo set the pressure downstream of the secondary source to zero
22 The influence of reflections from the primary source Secondary sourceFor a primary source next to the reflecting surface (D=0)Now, if R=1, thenThus the secondary source strength required to cancel the sound fieldbecomes infinite when
23 Adaptation in Feedforward Control Active Control of Transformer Noise, Conover 1956TRANSFORMERAMPLIFIERPHASEANGLEAMPLI-TUDEHARMONIC SOURCESOUND ANALYZERSOUND LEVEL METERLOUDSPEAKERMICROPHONEAn error microphone is introduced to monitor the performance.Changes in the disturbance and plant response, from loudspeaker to the microphone, require adaptation of the feedforward controller.
24 Single channel feedforward control PeriodicPrimarysourceErrorsensorElectricalreferencesignalSecondarysourceElectroniccontroller(Unaffected bysecondary source)ReferencesignalElectroniccontrollerElectroacousticsystemErrorPrimarycontribution
25 Single channel feedforward control ReferencesignalElectroniccontrollerElectroacousticsystemErrorPrimarycontributionAt the n-th harmonic the error signal can be completely cancelledifReference signal is
26 Control of random noise in a duct Sound fromPrimary sourceSecondarysourceErrorsensorDetectionsensorElectroniccontrollerThere are two main differences between the control of random andharmonic disturbancesThe detected signal x(t) is generally influenced by theelectroacoustics of the feedback path2. There is a constraint of causality on the controller
27 Control of random noise in a duct Measurement noiseat detection sensorPrimary pathSignal atdetection sensorSignal tosecondary sourceControllerErrorsignalError pathSignal dueto primarysourceFeedback pathMeasurement noiseat detection sensor
28 Optimal controller The block diagram becomes disturbance andmeasurementnoiseThe block diagram becomesPrimary andmeasurementnoiseErrorsignalController andfeedback pathError pathSince the system is linear and time-invariant, we can transposethe signal paths to giveController andfeedback pathError pathErrorsignalFiltered reference signalFiltered reference signal
29 Optimal controller Power spectral density of the error signal is where E[ ] is the expectation operator and * denote complex conjugationNowSoThis can be written in standard Hermitian quadratic form as(dropping the explicit dependence on frequency)
30 Optimal controllerThe power spectral density of the error signal can be written asGlobal minimumSo
31 Optimal controller To find minimum error substitute into To give which can be written asNowandCoherence between signals fromdetection sensor and error sensorprior to controlSoThe maximum possible attenuation in dB at each frequency is thus given by
32 Optimal controller So the optimal controller is given by Controller ErrorsignalError pathFeedback pathSo the optimal controller is given by
33 Digital implementation of the controller Sound fromPrimary sourceSecondarysourceErrorsensorDetectionsensorElectroniccontrollerDigital filterAnalogueto digitalconverterADCDigital toanalogueanti aliasreconstructionfilter
34 Digital implementation of the controller The overall frequency response of the controller isSampling timeFrequency response offilters and data convertersDigital filterCausality conditionThe controller must have a delay of secondsApproximate delay through an analogue filter is roughly due to 45°phase lag or 1/8 cycle of delay at its cut-off frequency, fcTotal delay through two filters which have a total of n poles is n/8fcThe cut-off frequency is typically 1/3 the sampling frequency (fs=1/T), so that fc=fs/3=1/(3T)Allowing 1 sample delay for the data converters and the digital filtermeans the total delay is given by
35 Causality condition - example Sound fromPrimary sourceSecondarysourceErrorsensorDetectionsensorElectroniccontrollerRectangular duct with largest dimension D=0.5m – single channelcontrol can only be achieved below about 300 HzSampling frequency = 1kHz (T=1ms)Two 4th order analogue filters (n=8)Delay in analogue path is about 4ms
36 Active control of sound in a duct – experimental work (Roure 1985) Side viewPlan view
37 Active control of sound in a duct – experimental work (Roure 1985) Amplitude spectra of the fan noise at the error microphone with a meanduct velocity of 9m/sActive control offdBActive control onFrequency (Hz)
38 Active control of sound in enclosures Electronic Sound AbsorberH.F. Olson and E.G. May, Journal of the Acoustical Society of America,pp , 1953
39 Active Control of Sound inside Cars Low-frequency engine noise in the car cabin can be controlled with 4 loudspeakers, also used for audio, and 8 microphones, also used for hands-free communication (Elliott et al. 1986).
41 Measured Results in a Demonstration Vehicle A-weighted sound pressure level at engine firing frequency
42 Active Sound Control in Propeller Aircraft System is standard fit on Dash 8 Q400 (Stothers et al. 2002)
43 Active Sound Control in Propeller Aircraft Periodic excitation generates intense harmonic soundfield inside cabin
44 Active Sound Control in Propeller Aircraft Spectrum of Pressure Inside Propeller AircraftDash-8 Series 200: Reduction 11.3 dB(L), 8.2 dB(A)Frequency (Hz)dB(A) re arbitrary level
45 Active Sound Control in Propeller Aircraft Control System for Propeller Aircraft Active Noise SystemCentralised digital system made by Ultra Electronics controls 5 harmonics with 48 structural actuators at 72 acoustic sensors, distributed throughout cabin.
46 Active Sound Control in Propeller Aircraft Typical Performance of an Active Aircraft SystemSYSTEM OFFSYSTEM ONSingle multichannel centralised digital controller used with 48 actuators and 72 sensors distributed throughout the cabin
47 Feedback control of Sound Active Headset using Feedback ControlIf no external reference signal is available, conventional feedback control can be used to control sound at low frequencies.
48 Feedback control of Sound Active Headset using Feedback ControlActive control offdBActive control onFrequency (Hz)
49 Feedback control of Sound Active Headset using Feedback Control
56 Active Control Performance (simulations) Feedback gainSound transmission ratio (dB)Integrated from 0-1kHzPiezoceramic ActuatorsForce ActuatorsFrequency (Hz)Sound transmission ratio (dB)Increasing gain
57 What Happens to the Panel Vibration? Integrated from 0-1kHzKinetic energy (dB)Increasing gainFrequency (Hz)Piezoceramic ActuatorsKinetic energy (dB)Force ActuatorsFeedback gain
58 Experimental Result (after Bianchi et al) Pressure (dB re arbitrary units)Gain limited by accelerometer resonanceCompensator used in feedback circuit
59 Concluding RemarksActive sound control is being used as an alternative to passivecontrol in many different applications especially at lowfrequenciesductsaircraftautomobileCombination of acoustic and vibration control maybe seen inthe future