Variable gain active noise cancellation system with improved residual noise sensing

0001The present invention relates to an active noise cancellation headset comprising a feedback loop including an earpiece, a noise sensor mounted in the earpiece, a processor and an actuator mounted in the earpiece, the processor being responsive to the noise sensor to provide drive signals to the actuator so as to effect noise cancellation in a region, and control means for controlling the loop so as to avert undesirable instability therein.

0002Conventionally, passive headsets and over-the-ear earplugs comprise a pair of earpieces coupled by a resilient headband. An annular foam pad attached to each earpiece forms a cushion between the shell of the earpiece and the user's head. The resilient headband presses the earpieces against the user's head. Ambient sound is attenuated before it reaches the wearer's ear by occlusion of sound by the earpieces and absorption of transmitted sound by materials within the earpieces. The degree of attenuation achieved depends upon the nature of the ambient noise and the qualities and characteristics of the individual headset or earplugs.

0003In various applications, however, passive attenuation is insufficient. Some environments are simply too noisy for comfort or even safety with only passive earplugs. In other environments, the elimination of extraneous noise is paramount, and satisfactory results cannot be achieved using passive means. Although the amplitude of the extraneous noise may be significantly diminished, it is almost impossible to completely isolate the wearer from extraneous noise wing passive means. In addition, passive earplugs attenuate all sound, regardless of whether the wearer needs or wants to hear particular sounds.

0004Active noise cancellation systems eliminate unwanted sound using destructive interference. Cancellation is achieved by propagating anti-noise, identical to the unwanted soundwaves but inverted, which interacts with the unwanted waveform and results in cancellation. A feedback active cancellation headset typically includes a sound generator in each earpiece for producing anti-noise, and a residual microphone, also located in each earpiece, to provide feedback signals to a controller which generates the proper anti-noise signals. Each microphone detects the unwanted noise within each earpiece and provides corresponding signals to the controller. The controller supplies anti-noise signals to the sound generator corresponding to the noise detected in the earpieces, but inverted, with respect to the unwanted waveform. When the anti-noise interacts with the noise within each earpiece, destructive interference between the noise and the anti-noise cancels the unwanted sound.

0005Ideally, the residual microphone in feedback systems perceives the same sounds as the eardrum of the listener. In this regard, effective proximity to the eardrum is vital; the goal of the cancellation systems is to reduce the unwanted noise at the eardrum to zero, but in fact operates upon the noise detected by the microphone. Consequently, it is desirable that the microphone be placed sufficiently dose to the eardrum to detect a reasonably similar noise field to that perceived by the listener. The eardrum, however, is located deep within the ear canal. Placing a microphone within the ear canal is generally impractical and very uncomfortable for the user. In addition, locating the microphone a significant distance away from the sound generator introduces a phase shift between the residual and cancellation signals and causes instability. As a result, active cancellation systems conventionally approximate the sound perceived by the listener by locating the microphone as close to the ear canal as possible without actually penetrating it.

0006A problem associated with feedback cancellation systems is that they are prone to instability. Feedback systems tend to become unstable, for example, if the bandwidth of the system is too broad or the gain of the system is too high. When instability occurs, the system usually emits a loud noise that is generally unpleasant and occasionally dangerous. Consequently, the maximum range and effectiveness of feedback systems are limited by parameters designed to keep the feedback system stable.

0007This problem is addressed in one manner in EP-A-0 212 840. However, the disclosed system relies on analysing the residual signal which is a function of noise amplitude as well as control loop transfer function.

0008To effect maximum cancellation, the waveform of the interacting anti-noise should exactly match the unwanted waveform, but should be inverted. The acoustic properties of each earpiece, however, affect the characteristics of the anti-noise waveform. The effect of the acoustic properties may be corrected by processing the residual signal according to a transfer function characteristic of the acoustic properties of the system to compensate for the effects. However, these acoustic properties of the headset are not constant under all conditions, and may vary with the force applied to the earpiece onto the user's head. When high pressure is applied to the earpiece, or when the headset is removed from the user's head, the variation of the earpiece's acoustic properties, particularly the volume and acoustic resistance, may cause instability in the feedback loop. This instability, in turn, causes the control loop to generate unstable oscillations, producing unpleasant and potentially even harmful noise.

0009A headset according to the present invention is characterised in that the control means includes pressure sensing means for detecting the pressure of the earpiece against a user's head and controls the transfer function of the loop in dependence on the output of the pressure sensing means so as to avert said undesirable instability.

0010The earpiece may include first and second portions, moveable relative to each other, and the pressure sensing means may comprise a switch operated by relative movement of said first and second portions, the control means being operative to reduce the gain of the loop when the switch changes state at a first threshold pressure due to an increase in pressure between the earpiece and a user's head. It is then preferable that the pressure sensing means comprises a further switch operated by relative movement of said first and second portions, the control means being operative to reduce the gain of the loop when the further switch changes state at a second threshold pressure due to a decrease in pressure between the earpiece and a user's head and the second threshold being lower than the first threshold. The switch and or the further switch may be in a power supply path to the processor and, when open, set the gain of the loop to zero by removing power from the processor.

0011The pressure sensing means may optionally comprise a pressure sensitive resistor.

0012The pressure sensing means may also optionally comprise a subsonic signal generator for producing a subsonic frequency signal and adding the subsonic frequency signal to said drive signals so that the actuator is driven by the drive signals and the subsonic frequency electrical signal, extraction means for extracting the subsonic frequency signal dependent portion of the microphone's output signal and agc means for adjusting the gain of the loop independence on the amplitude of the extracted subsonic frequency signal dependent portion of the microphone's output signal. Preferably, the extraction means comprises a low-pass filter, a band-pass filter or a synchronous detector.

0013Preferably, the agc means is operative to vary the gain of the loop so as to tend to keep constant the amplitude of the extracted subsonic frequency signal.

0014Embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which:- <ul id="ul0001" list-style="none" compact="compact"><li>Figure 1 illustrates a noise cancelling headset according to the present invention;</li><li>Figure 2 is a sectional view of a headset incorporating a pressure sensitive noise cancellation system according to various aspects of the present invention;</li><li>Figure 3 is a schematic diagram of a switching system according to various aspects of the present invention;</li><li>Figure 4 is a schematic diagram of a pressure sensitive noise cancellation control system according to various aspects of the present invention;</li><li>Figures 5 and 6 depict front and sectional side views, respectively, of a switched headset according to various aspects of the present invention;</li><li>Figures 7A-B are schematic diagrams of variable gain and pressure sensitive noise cancellation control systems according to various aspects of the present invention; and</li><li>Figure 8 is a block diagram of a headset noise cancellation system using a subsonic acoustic signal according to various aspects of the present invention.</li></ul>

0015The term headset as used herein includes but is not limited to ear defenders, headphones, earphones and telephone headsets and handsets. The term noise as used herein includes both periodic and non-periodic acoustic signals, including vibrations in solids and fluids.

0016Referring to Figure 1, a noise cancellation headset 100 suitably comprises a first earpiece 1 and a second earpiece 2, coupled by a resilient headband 3. External microphones 300a, b are mounted on the outside of earpieces 1, 2, suitably beneath headband 3. Earpieces 1 and 2 enclose respective sound generating units and internal microphones (not shown), which are electrically coupled by respective groups of wires 4, 5 to a controller 6. Controller 6 may be formed integrally with one or more of earpieces 1 and 2, headband 3 or otherwise one or more of the other components to form a unitary headset. Preferably, however, controller 6 is constituted as a separate unit to facilitate assembly, and, where batteries are employed as a power source, provide adequate space to accommodate the batteries.

0017Wires 4 and 5 are preferably removably coupled to controller 6 through a suitable connector 5a. If desired, controller 6 may also be connected, suitably through a conventional 3.5mm (1/8 inch) audio jack 56, to a source of sound desired to be passed through headset 100 and communicated to the user. Exemplary pass through sound sources include such things as conventional consumer audio equipment such as radios, and tape or CD players, and intercom or other communication system.

0018A separate controller 6 facilitates incorporating controller 6 into another piece of equipment or fixture, such as, for example, into the armrest or headrest of an aircraft seat (generally indicated in Figure 1 as 6a), where desired. Use of a separate controller unit also makes the portion of the system worn on the user's head lighter and more comfortable. Further, removable portion of the system is less expensive. In contexts where headset units are issued at various times to numerous different users for use with a given installed controller, such as in airline passenger use, this can be a significant advantage. In the context of airline use, the conventional aircraft audio signals would be the pass through signal source to controller 6.

0019Controller 6 may comprise a virtual earth controller. Examples are described in EP-A-0 596 971. One aspect of a controller 6 such as described in EP-A-0 596 971 is particularly adapted to canceling harmonically related tones in noise. Consequently, such a virtual earth controller 6 employs a synchronizing signal input, suitably derived from an appropriate synchronizing signal source, such as, for example, from a phase locked loop circuit fed with a signal from a microphone mounted outside the earpieces.

0020It is recognized that changes in the pressure of earpiece 1 against the user's head affect the acoustic properties of earpiece 1 and may cause instability. Instability can therefore be mitigated by varying the gain of the feedback loop according to the sensed pressure. To this end, when pressure is applied or released to the headset, gain may be reduced or the cancellation system disabled altogether.

0021Referring to Figure 2, an earpiece 400 incorporating this feature preferably incorporates a mechanism for sensing when the pressure of the earpiece against the user's ear is outside of a predetermined range, for example, exceeding a maximum value, or less than a minimum value. In the embodiment of Figure 2, such an earpiece 400 comprises a first shell portion 407, a second shell portion 410, and annular foam pad 30. First shell portion 407 is generally semi-spherical or cylindrical, and concave. An annular wall 408 extends from within first shell portion 407, generally co-axial, with shell portion side wall, to a region beyond the plane of the shell portion lip. A flange 409 extends radially outwardly from the free axial extremity of annular wall 408. Second shell portion 410 comprises an annulus 410a, first and second flanges 410b and 410c. Annulus 410a suitably manifests an outer radius approximately equal to the radius of the lip of shell portion 407, and is disposed coaxially with first shell portion 407. First flange 410b extends radially inwardly from an axial extremity of annulus 4 10a between flange 409 and the lip of first shell portion. Second flange 410c extends radially inward from the other axial extremity of annulus 410a. Foam pad 30 is affixed coaxially to the outer surface of second flange 410b of the second shell portion 410.

0022Sound generator 22 is suitably mounted coaxially within annular wall 408 and microphone 28 is mounted on the open side of sound generator 22. First and second sets of switches 414, 416 are mounted on the respective sides of flange 409. Switches 414, 416 may be any suitable state changing switch normally closed or normally open depending upon the particular design of the cooperating circuitry. Switches 414, 416 are designed to change state upon application of a predetermined amount of pressure. In the preferred embodiment, switches 414, 416 are momentary contact membrane switches. Membrane switches are particularly suited to the disclosed embodiment because they are thin, easily assembled and inexpensive. Switches 414, 416 in the preferred embodiment require little compression to change state, and preferably include or cooperate with circuitry to induce hysteresis and prevent switches 414, 416 from repeatedly switching state at pressures fluctuating near the threshold pressure. The hysteresis circuitry may be external or incorporated into the switches. Although shown in Figure 2 as each including two switches, switch sets 414 and 416 may include any suitable number of switches. For example, three switches, spaced at approximately 120 degree intervals around flange 409, may be used to effectively detect pressure upon most portions of earpiece 1. Alternatively, however, a single switch may be used in each set if the headset design is particularly sensitive to pressure at a certain point or from a certain angle.

0023First switches 414 are mounted on flange 409, directed towards second flange 410c of second shell portion 410. A predetermined number, e.g. three, of elastomeric blocks 415, are affixed between flange 409 and second flange 410c of the second shell portion 410, preferably angularly spaced by about 120°. Second switches 416 are located on flange 409, directed towards first flange 410b of second shell portion 410.

0024The disposition of flange 409 of first shell portion 407 relative to first flange 410b and second flange 410c is a function of the pressure of the earpiece against the user's head. Switches 414 and 416 are disposed for actuation in response to contact with flanges 410b and 410c of second shell portion 410, respectively. The characteristics of elastomeric blocks 415 and the distance between flanges 410b and 410c are chosen to cause contact between switches 414 and flange 410c and switches 416 and flange 410b in a response to predetermined levels of pressure. For example, in the absence of pressure, elastomeric blocks 415 may force switches 416 into contact with flange.410b. When the pressure exceeds a predetermined lower threshold, compressing clastomeric blocks.415, switches 416 move out of contact with flange 410b, (and hence assume their normal condition). When the pressure exceeds a predetermined upper threshold sufficient to compress the elastomeric blocks 415 by a predetermined amount, switches 414 come into contact with flange 410c and are actuated.

0025If desired, switches 414 and 416 may be employed to deactivate controller 6 if the pressure of the earpiece against the user's head is outside of a predetermined range, so that controller 6 is deactivated when the pressure is less than a predetermined minimum value, e.g. when the earpieces are not being worn, or when the pressure exceeds a predetermined level, rendering the system susceptible to oscillations.

0026Referring to Figure 3, such an embodiment may be implemented employing normally closed switches 414 and 416 serially interposed between controller 6 and its power source. Controller 6 is suitably supplied with power from a battery 517. The current path from the positive terminal of battery 517 to controller 6 passes through the normally closed first and second switches 414, 416. Only single first and second switches 414, 416 are shown in Figure 3 for clarity. In the exemplary embodiment, all of the switches are in a series configuration with each other.

0027Referring to Figures 1, 2 and 3, before earpiece 1 is placed against the user's head, elastomeric blocks 415 force second shell portion 410 away from first shell portion 407. This causes first flange 410b of second shell portion 410 to contact second switches 416 causing them to be open. When one of second switches 416 is open, current cannot flow to controller 6 from battery 517.

0028When a user puts on headset 100, first earpiece 1 is pressed against the user's head by the resilience of headband 3. The force applied by headband 3 to first shell portion 407 causes elastomeric blocks 415 to be compressed between first and second shell portions 407,410. As elastomeric blocks 415 are compressed, first flange 410b of second shell portion 410 moves away from second switches 416, allowing them to close. At this point, both first and second switches 414, 416 are closed, allowing the supply of current flow from battery 517 to controller 6. Controller 6 then drives sound generator 22 to effect noise cancellation at the user's ear.

0029If, for some reason, an additional force is applied to force earpiece 1 towards the user's head, elastomeric blocks 415 are further compressed and first switches 414 are caused to open by coming into contact with second flange 410c of second shell portion 410. The opening of anyone of first switches 414 interrupts the flow of current from battery 517 to controller 6, deactivating controller 6. Since controller 6 has been deactivated, the feedback loop is unable to go into oscillation so that the risk of damage to the user's ear is avoided.

0030As an alternative to deactivating controller 6, the gain of the feedback loop may be varied in accordance with the actuation of switches 414 and 416 to preclude oscillation. Referring to Figure 4, a suitable circuit using normally open switches 414a and 416a for effecting such gain control comprises an amplifier 618, including a feedback network, respective resistors 620, 622 and 623, and a field effect transistor (FET) 621. Amplifier 618 is provided in the feedback path of controller 6. The gain of amplifier 618 is set by feedback network 619. Resistor 620 is selectively coupled into feedback network 619, in accordance with the state of switches 414a and 416a, to selectively vary the gain of amplifier 618. Resistor 620 is coupled between a node of feedback network 619 and the drain of FET 621. Switches 414a and 416a are included in the biasing circuit of FET 621. In this embodiment, first and second switches 414a, 416a are normally open, momentary contact switches, suitably membrane switches. FET 621 is selectively rendered conductive or non-conductive accordingly to the state of switches 414a and 416a. The source of FET 621 is connected to a reference ground potential. Each switch of first and second sets of switches 414a and 416a (only one of each shown for simplicity) are connected in parallel between one terminal of resistor 622 and gate of FET 621. The other terminal of resistor 622 is coupled to a reference voltage rail V_ref which is negative with respect to the reference ground potential. The gate of FET 621 is also coupled to a positive supply rail + V via resistor 622.

0031Referring to Figures 2 and 4, before a user puts on headset 100, second switches 416a are closed. This causes the potential at the gate of FET 621 to be reduced, turning off FET 621 and thereby reducing the gain of amplifier 618. Under normal wearing conditions, first and second switches 414, 416a are both open, FET 621 is turned on and the gain of amplifier 618 is set by the feedback network 619. If an additional force is applied to earpiece 1, first switches 416 close, turning off FET 621 and thereby reducing the gain of amplifier 618 by effectively removing resistor 620 from feedback network 619. The reduction in the gain of amplifier 618 ensures that stability is maintained regardless of the transfer function of the acoustic component of the feedback loop.

0032Alternatively, the embodiments shown in Figures 3 and 4 may be combined. For instance, second switches 416 may cut the supply of current to controller 6 when headset 100 is removed. First switches 414a, on the other hand, as in Figure 4, may reduce the gain of amplifier 618 when earpiece 1 is pressed too tightly to the user's head, but continue to supply power to headset 100.

0033The gain of the feedback loop may also be varied as a function of the pressure of earpiece 1 against the user's head. For example, suitable pressure transducers, such as an element having a resistance which varies according to pressure or other pressure sensitive element, maybe used to affect the gain of the headset circuit according to the force applied to earpiece 1 against the user's head. Referring to Figures 5 and 6, an earpiece 1 employing resistive pressure transducers comprises a shell 700, sound generator 22, microphone 28, an annular plate 845, pressure transducers 847, and annular foam pad 30. Shell 700 supports sound generator 22 coaxially within a central chamber 832 formed by shell 700. If desired, shell 700 may include a transverse sidewall 700a, inset from the outer periphery of the shell (e.g., adjacent pad 30). Apertures communicating with chamber 832 to the rear of sound generator 22 may be formed through transverse wall 700a to enhance the low frequency response of the earpiece 1. Disposition of the passageways in the sidewall reduces susceptibility to blockage.

0034Chamber 832 is open at the earside of shell 700. Microphone 28 is supported at the open end of chamber 832. Annular plate 845 is located coaxially with shell 700 and separated from shell 700 by a predetermined number, e.g. three of resistance pressure transducers 847, preferably disposed with equal angular spacing about plate 845, e.g. separated by 120°. Transducers 847 are electrically connected in series. Any numbers of transducers may be used, although three are shown in Figure 5. An annular seal 849 forms a circumferential wall to the chamber defined between annular plate 845 and shell 700. Annular foam pad 30 is affixed coaxially to the outer face of annular plate 845.

0035Referring to Figure 7a, an embodiment of a circuit according to the present invention is shown in which the gain of controller 6 may adjust according to the force exerted on earpiece 1 by, e.g. the user's head. Suitably, a voltage divider is formed by pressure transducers 947 connected in series with a resistor 950. The output from the voltage divider is suitably amplified by an inverting amplifier 951, the output of which is applied to the gate of FET 621. The voltage existing at the voltage divider is proportional to the pressure exerted upon pressure transducers 947. As the pressure upon pressure transducers 947 increases, the gate voltage for FET 621 decreases, reducing the conductivity of FET 621 and accordingly reducing the gain of feedback amplifier 618.

0036Alternatively, where feedback amplifier 618 includes multiple stages, the coefficients of filters included in feedback amplifier 618 may be varied as the pressure increases to maintain the frequency response of the cancellation system loop. Varying the filter coefficients of the cancellation system preserves the stability of the system and maintains the bandwidth. Thus, increased pressure of earpiece 1 upon the user's head varies the gain and/or frequency response of the cancellation system to avoid instability.

0037A similar gain control system applied to a feedforward system enhances the cancellation system's effectiveness. When pressure is sensed upon earpiece 1, the coefficients of the filters are adjusted to vary the gain, frequency response, and phase of the feedforward cancellation signal. With proper calibration, the proper amplitude and phase relationship of the cancellation signal to the primary signal may be maintained.

0038In a further embodiment, controller 6 may be deactivated when the pressure of earpiece 1 against the user's head falls below a predetermined threshold. Referring to Figure 7b, the output from the voltage divider circuit and amplifier 951 may be suitably applied to the input of a suitable threshold comparator circuit, such as a Schmitt trigger circuit 952. The power supply is permanently connected to amplifier 951, Schmitt trigger 952 and the voltage divider to provide continuous power. The output of the Schmitt trigger circuit 952 is connected to the base of a switching transistor 953 connected in series with battery 517 and controller 6. When the output of the pressure-varying voltage divider and amplifier 951 falls below a lower threshold voltage of Schmitt trigger 952, transistor 953 turns off, opening the circuit connecting controller 6 to battery 517. Consequently, the cancellation system is turned off until the pressure on transducers 947, and correspondingly the voltage provided to the Schmitt trigger, increases enough to switch the output of Schmitt trigger 952 on and reconnect battery 517 to controller 6.

0039A wide range of other techniques for measuring the pressure of earpiece 1 against the user's head may also be employed. For example, a subsonic acoustic signal may be employed to determine the current acoustic properties of the earpiece. Referring to Figure 8, a sound generating unit 22 and a microphone 28 are mounted within an earpiece 1. A feedback loop is formed between microphone 28 and sound generating unit 22 through a cancellation circuit 310. A summing circuit 1065 adds the output of a subsonic frequency signal generator 1066 to the output of cancellation circuit 310. The subsonic component of the output of sound generating unit 22 is, of course, outside of the normal human audible range and is thus inaudible to a person using earpiece 1. However, the subsonic acoustic signal, generated within earpiece 1, is detected by microphone 28. The level of the subsonic acoustic signal detected by microphone 28 is dependent upon the acoustic properties of the earpiece 1, which are affected by the pressure between earpiece 1 and the user's head.

0040The subsonic signal may be mixed with the cancellation signal generated by cancellation circuit 310 in a feedforward configuration so that the subsonic signal remains at a constant level regardless of the gain of the system. This level may be set for the particular cancellation system so that cancellation circuit provides the same level of cancellation regardless of the individual characteristics of sound generator 22, microphone 28 and the other system components. This arangement has particular application to systems in which sound generator 22 and microphone 28 are interchangeable with others, such as airplane headsets. Alternatively, subsonic signal may be mixed with the residual signal generated by microphone 28, so that the amplitude of the subsonic signal is modulated by the gain of the system. This forms a feedback loop in which the system continuously adjusts the gain to maintain the subsonic signal at a particular level.

0041The output of microphone 28 is passed through a low pass filter 1068 to a suitable level detecting circuit 1069, e.g. a rectifying and smoothing circuit. Alternatively, filter 1068 may be a bandpass filter having an appropriate centre frequency and quality. Subsonic frequency signals are transmitted by filter 1068, and the gain or frequency response of cancellation circuit 310 is adjusted according to the output of level detecting circuit 1069, which varies according to the pressure of earpiece 1 against the user's head. Similarly, level detecting circuit 1069 may replace the gain control switch circuit in Figure 4, the output of level detecting circuit would be amplified and applied to the gate of FET 621. Like the switch circuit of Figure 4, the output of level detecting circuit 1069 reduces amplifier's 618 gain by turning FET 621 on when sufficient pressure is exerted upon earpiece 1. Alternatively, the output of level detecting circuit 1069 may vary the frequency response of the cancellation system according to the pressure exerted on earpiece 1. Similarly, the output of level detector 1069 may be used to determine when headset 100 has been removed from the user's head.