System and method for protecting loudspeakers using overexcursion, frequency compensation, and nonlinearity correction

Abstract

In at least one embodiment, an audio amplifier system is provided. The system includes a speaker and an audio amplifier. The speaker transmits an audio output into a listening environment. The audio amplifier is programmed to receive an audio input signal and, based on the audio input signal, generate an excursion signal corresponding to a first excursion level of a voice coil. The audio amplifier is further programmed to limit the excursion signal to a maximum excursion level and determine a target pressure for a housing of the speaker based on the maximum excursion level. The audio amplifier is further programmed to generate a target current signal based on at least the target pressure and convert the target current signal into a target voltage signal to a target drive signal to drive the voice coil to the maximum excursion level.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. Provisional Application Serial No. 62 / 955,138, filed December 30, 2019, the disclosure of which is incorporated herein by reference in its entirety.

[0003] This application is generally related to U.S. application serial number 62 / 955,125, filed on December 30, 2019, entitled “SYSTEM AND METHOD FORADAPTIVE CONTROL OF ONLINE EXTRACTION OF LOUDSPEAKER PARAMETERS,” the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0004] One or more aspects disclosed herein generally relate to systems and methods for providing advanced loudspeaker protection utilizing over-excursion, frequency compensation, and nonlinearity correction. For example, the aspects disclosed herein may correspond to, but are not limited to, precise over-excursion compression and limiting, frequency compensation, and nonlinearity correction for passive radiator, vented sealed box, or infinite baffle dynamic acoustic transducer loudspeaker combinations. These may be suitable for systems independent of look-ahead implementations such as active noise cancellation (ANC), and may be suitable for or implemented for adaptive or automatic tuning for use with various amplifier topologies. These aspects will be discussed in more detail below. Background Art

[0005] U.S. Patent No. 10,667,040 (the “040 patent”) issued to French provides an audio amplifier system including a memory and an audio amplifier. The audio amplifier includes the memory and is programmed to receive an audio input signal and generate a target current signal based on the audio input signal and a velocity of a diaphragm of a loudspeaker. The audio amplifier is further programmed to generate a correction current signal based on at least the target current signal and a predicted position of a voice coil of the loudspeaker, and to determine the predicted position of the voice coil of the loudspeaker based on a flux density value. The flux density value corresponds to the product of the magnetic flux of an air gap of the voice coil in the loudspeaker and the length of the voice coil wire in the loudspeaker. Summary of the Invention

[0006] In at least one embodiment, an audio amplifier system is provided. The system includes a speaker and an audio amplifier. The speaker includes a voice coil for generating an audio output into a listening environment. The audio amplifier is operably coupled to the speaker and is programmed to receive an audio input signal and, based on the audio input signal, generate an excursion signal corresponding to a first excursion level of the voice coil. The audio amplifier is further programmed to limit the excursion signal to a maximum excursion level and determine a target pressure for a housing of the speaker based on the maximum excursion level. The audio amplifier is further programmed to generate a target current signal based on at least the target pressure and convert the target current signal into a target voltage signal to a target drive signal to drive the voice coil to the maximum excursion level.

[0007] In at least another embodiment, a computer program product embodied in a non-transitory computer-readable medium is provided, the computer program product being programmed to protect a loudspeaker. The computer program product includes instructions for receiving an audio input signal and generating an excursion signal corresponding to a first excursion level of a voice coil based on the audio input signal. The computer program product also includes instructions for limiting the excursion signal to a maximum excursion level and determining a target pressure for a housing of the loudspeaker based on the maximum excursion level. The computer program product also includes instructions for generating a target current signal based at least on the target pressure and converting the target current signal into a target voltage signal to a target drive signal to drive the voice coil to the maximum excursion level.

[0008] In at least one embodiment, a method for protecting a loudspeaker is provided. The method includes receiving an audio input signal and generating an excursion signal corresponding to a first excursion level of a voice coil based on the audio input signal. The method also includes limiting the excursion signal to a maximum excursion level and determining a target pressure for a housing of the loudspeaker based on the maximum excursion level. The method also includes generating a target current signal based at least on the target pressure and converting the target current signal into a target drive signal to drive the voice coil to the maximum excursion level. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The embodiments of the present disclosure are particularly pointed out in the appended claims. However, other features of the various embodiments will become more apparent and will be best understood by referring to the following detailed description taken in conjunction with the accompanying drawings, in which:

[0010] Figure 1 Generally depicts an example of a closed loudspeaker system;

[0011] Figure 2 The general description includes various aspects of the transducer;

[0012] Figure 3 A general description includes various aspects of passive radiators;

[0013] Figure 4 A model generally showing the elements of a loudspeaker system associated with the transducer and passive radiator;

[0014] Figure 5 1. A system for estimating Kms(x) and Rms(x) in a speaker system is generally shown according to one embodiment;

[0015] Figure 6 generally illustrates an amplifier system for correcting distortion in a loudspeaker system according to one embodiment;

[0016] Figure 7 According to one embodiment Figure 6 The amplifier system further includes a core correction block;

[0017] Figure 8 depicts a correction system used as a voltage source to drive a voice coil according to one embodiment;

[0018] Figure 9 depicts a system for providing advanced loudspeaker protection according to one embodiment;

[0019] Figure 10 corresponds to a graph showing the behavior of a compressor and limiter and a loudspeaker according to one embodiment;

[0020] Figure 11 corresponds to a graph showing a slow attack that allows for larger over-excursions in addition to allowing low-frequency artifacts in order to avoid over-compression;

[0021] Figure 12 corresponds to a graph showing a fast attack for avoiding low frequency artifacts but allowing for excessive excursion;

[0022] Figure 13 corresponds to a graph depicting the effect of a limiter controlling the maximum position in the absence of a compressor;

[0023] Figure 14 Depicting a system for protecting a loudspeaker from an over-temperature condition of a voice coil, according to one embodiment;

[0024] Figure 15 Describes a system for providing accuracy of indirectly measurable temperature of a voice coil according to one embodiment; and

[0025] Figure 16A method for providing advanced loudspeaker protection according to one embodiment is described. DETAILED DESCRIPTION

[0026] Detailed embodiments of the present invention are disclosed herein as required; however, it will be understood that the disclosed embodiments are merely exemplary of the invention that may be embodied in various and alternative forms. The drawings are not necessarily drawn to scale; some features may be exaggerated or minimized to illustrate details of particular components. Therefore, the specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to variously employ the present invention.

[0027] It should be appreciated that the controllers disclosed herein may include various microprocessors, integrated circuits, memory devices (e.g., FLASH, random access memory (RAM), read-only memory (ROM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or other suitable variants thereof) and software that cooperate with each other to perform the operations disclosed herein. In addition, such controllers disclosed utilize one or more microprocessors to execute computer programs embodied in non-transitory computer-readable media that are programmed to perform any number of the disclosed functions. In addition, the controllers provided herein include a housing and various numbers of microprocessors, integrated circuits, and memory devices (e.g., FLASH, random access memory (RAM), read-only memory (ROM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM)) positioned within the housing. One or more controllers as disclosed also include hardware-based inputs and outputs for receiving data from and transmitting data to other hardware-based devices as discussed herein, respectively.

[0028] When dynamic transducers (or dynamic speakers) increase their acoustic output, they also increase their distortion. This fundamental relationship drives the transducer's size, weight, cost, and inefficiency, all of which are undesirable. This is particularly true for transducers used in automotive applications, where these performance issues are significant. At the same time, there is a growing need for higher-output, lower-distortion systems that can achieve or provide the desired active noise cancellation (ANC), engine-order noise cancellation (EOC), independent sound zones (ISZ), and echo cancellation for voice recognition.

[0029] Therefore, there are current sensing methods, such as those described by Klippel, which attempt to minimize the distortion of the transducer through signal processing, which (if used properly) can in turn enable transducer designers to achieve smaller, lighter, lower cost or more efficient solutions based on the desired trade-offs. However, these methods can be computationally expensive (e.g., 100 million instructions per second (MIPS) or more), especially in multi-channel applications (such as those found in automobiles). In addition, these methods typically require an embedded microcontroller and a digital signal processor (DSP). Therefore, there is a need for a low MIPs algorithm (e.g., which provides relatively low processing requirements) and a low hardware cost method for nonlinear distortion correction as provided herein. In addition, the solution should be compatible with automotive hardware that requires relatively low processing requirements.

[0030] In general, at a basic level, once nonlinearities in the transducer are actively controlled and / or corrected, transducer and system designers have flexibility in the tradeoffs that may be necessary in a loudspeaker. This can improve size, weight, cost, and efficiency based on design goals. For example, the embodiments disclosed herein can provide better control over the transducer's displacement or offset and voice coil current, which can allow the transducer to be driven closer to its limits and, therefore, provide greater output. Furthermore, the embodiments disclosed herein provide enhanced control over the nonlinear performance of the transducer and can enhance the performance of acoustic algorithms that depend on the transducer's linearity or response, such as ANC, RNC, EOC, ISZ, echo cancellation, and the like.

[0031] The embodiments disclosed herein can be: (i) robust and inherently predictable in terms of stability, repeatability, and auditability (i.e., not a black box), (ii) computationally simple, with low to very low MIPs, sensorless, (iii) adaptive with simple current sensing, and (iv) algorithmically simplified and operable in a DSP environment that may not require an accompanying embedded controller in order to be adaptive.

[0032] Figure 1 An example of a closed speaker system 100 according to one embodiment is generally depicted. System 100 includes a housing 101 that generally includes a speaker 102 (or transducer) (e.g., an active speaker or main driver) and a passive radiator 104 (or a passive radiating cone that does not receive electrical energy in the form of an audio input signal). Housing 101 generally represents a common speaker enclosure for transmitting audio signals, and aspects related to transducer 102 and passive radiator 104 will be discussed in more detail below.

[0033] Figure 2The general depiction includes various aspects of transducer 102. For example, transducer 102 generally includes a cone (or diaphragm) 110 and a voice coil 112. A surround (or suspension) 114 is attached to the end of diaphragm 110. A bobbin 116 surrounds voice coil 112 and is positioned within an air gap 118. An external magnet (or magnets) 120 surrounds air gap 118 and at least a portion of voice coil 112 and bobbin 116. A spider 122 surrounds a portion of bobbin 116.

[0034] Typically, an audio input signal corresponding to audio data is provided to the voice coil 112. The voice coil 112 and the magnet 120 are magnetically coupled to each other, and the audio input signal causes the diaphragm 110 to move linearly along a vertical axis based on the polarity of the audio input signal. The diaphragm 110 is generally flexible and deflects in two directions along the vertical axis in response to the magnetic field transmitted between the voice coil 112 and the magnet 120. The skeleton 116 is attached to the diaphragm 110 and undergoes similar displacement (or movement along the vertical axis) as the diaphragm 110. Due to the linear displacement of the diaphragm 110, the transducer (or speaker) 100 transmits the audio input signal into a room or other environment for consumption by a user. The centering blade 122 is generally configured to prevent the diaphragm 110 from moving horizontally during the linear movement of the diaphragm 110 in the vertical direction or axis.

[0035] Figure 3 A general depiction includes various aspects of the passive radiator 104. In general, the passive radiator 104 can include all of the mentioned components that make up the transducer 102, except for the voice coil 112 and the magnet 120. The passive radiator 104 can use the sound trapped within the housing 101 to generate resonance to provide low frequencies (i.e., bass). The passive radiator 104 can generate frequencies based on the mass and elasticity (or compliance) of the air within the housing 101. The passive radiator 104 can be tuned to the housing 101 by changing its total diaphragm mass (including the weight of the diaphragm 110 or cone). When the transducer 102 generates air pressure due to the linear displacement of the diaphragm 110, this air pressure causes the passive radiator 104 to move.

[0036] Figure 4 A model of the elements associated with the transducer 102 and the passive radiator 104 in the speaker system 100 is shown. Generally, by mathematically modeling the behavior of the voice coil 112 (or the moving coil of the transducer 102) and other mechanical elements in the speaker system 100, nonlinear behavior can be calculated and corrected for using amplifiers and real-time signal processing. These aspects will be discussed in more detail herein.

[0037] There are many ways to model a loudspeaker system. However, if the physical elements of the system are well understood up front, as is the case here, a model that fits those elements can be computationally simpler and most easily tuned. Aspects disclosed herein attempt to model the physical elements (e.g., transducer 102 and passive radiator 104) and their interactions in a loudspeaker system 100 in a way that is directly computable, adaptively tunable, and corrects for when the elements behave in nonlinear ways.

[0038] There are typically four subsystems in the speaker system 100: (1) a transducer 102 (which transduces an electrical signal from an amplifier (not shown) into a mechanical output (not shown) (e.g., the mechanical output can be considered to be motion, which in turn transduces the mechanical output into an acoustic signal); (2) a passive radiator 104 (which resonates with the housing 101 and transducer 102 to produce an acoustic output at a lower frequency); (3) the housing 101, which couples the passive radiator 104 to the transducer 102 (via pressure) and isolates the back pressure of both the passive radiator 104 and the transducer 102 from the front pressure; and (4) an amplifier and signal processing (not shown). Two simplified subsets of the speaker system 100 may also be used, such as a vented system in which the passive radiator 104 is replaced with an acoustic mass generated using a port in the housing 101, and a closed box system with a simple sealed housing without vents or passive radiator 104. Figure 4 Three mechanical subsystems are shown and are similar to a two-body resonant system.

[0039] In general, the mechanical elements of transducer 102 can be modeled as a spring with a stiffness (e.g., Kms_TD), damping (e.g., Rms_TD), and a moving mass (e.g., M_TD). M_TD corresponds to the mass of all moving parts (including the air coupled to diaphragm 110). Rms_TD corresponds to the friction loss of the combined surround 114 and centering support 122. Kms_TD corresponds to the spring stiffness of the combined surround 114 and centering support 122. In a similar manner, passive radiator 104 can be modeled as a stiffness (e.g., Kms_PR), damping (e.g., Rms_PR), and a moving mass (e.g., M_PR). Transducer 102 and passive radiator 104 can be considered as the two bodies of system 100. The force of the coupled body can be modeled by the pressure in housing 101 (e.g., relative to the ambient pressure outside housing 101) multiplied by the surface area (e.g., Sd_TD) of the diaphragm 110 of the transducer 102 and the surface area of ​​the diaphragm 110 of the passive radiator 104. The compressibility of the air in housing 101 can be modeled as a spring with a stiffness of kappa "κ" (i.e., the adiabatic index of air, approximately 1.4) multiplied by the box pressure.

[0040] In the case of the voice coil 112 (or the moving coil of the transducer 102 ), the driving force F_1 may be modeled by the magnetic field strength (e.g., “B”) in the air gap 118 multiplied by the length “L” of the conductor in the field multiplied by the current in the conductor (e.g., the voice coil 112 ).

[0041] F1(t)=B·L·i vc (t) = BL·i Equation (1)

[0042] Reference frame x1(t) is defined with respect to the position of diaphragm 110 of transducer 102. Similarly, reference frame x2(t) is defined with respect to the position of diaphragm 110 of passive radiator 104. The positive direction of x1(t) is defined as movement into housing 101, and the positive direction of x2(t) is defined as movement out of housing 101.

[0043] The force of a moving mass is the mass times the acceleration, the force of a spring is the distance from rest times the spring stiffness, and the force of friction (or damping) is the velocity times the friction.

[0044] The force on the moving mass of transducer 102 (eg, MmsTD) may be represented by the following equation:

[0045]

[0046] where x1(t) is shown as x1.

[0047] In a similar manner, the force on the moving mass of the passive radiator 104 can be represented by the following equation:

[0048]

[0049] where x2(t) is shown as x2.

[0050] Next, it may be necessary to calculate the pressure "p" based on the position of the diaphragm 110 of the transducer 102 and the diaphragm 110 of the passive radiator 104. This can be accomplished by first calculating the volume change of the housing 101 (e.g., Vol_1), which in turn can be the volume of the housing 101 (e.g., Vol_0) minus the volume occupied by the displacement of the diaphragm 110 of the transducer 102 and the passive radiator 104 from their rest position. It is known that the volume of air is proportional to the pressure, and therefore:

[0051] Vol(x1, x2)=Vol0+(s D_TD x1-S D_PR ·x2) Equation (4)

[0052] Next, by relating the relative pressure "p" in the housing to the relative volume and the pressure outside the housing (of the ambient environment), p_ambient, the new pressure caused by the volume change can be calculated by the following equation:

[0053]

[0054] Note that in the free body diagram (i.e. Figure 4 The “p” in equation (5) is p(x1, x2).

[0055] If Vol_0 is allowed to be the volume of the housing 101 when the diaphragms 110 (of both the transducer 102 and the passive radiator 104) are at rest, the change in pressure relative to the ambient pressure can be shown by Equation 6 as shown below.

[0056] By combining equations (4) and (5) to calculate the pressure in the housing 101 relative to the surrounding environment based on X1 and X2, the following equation is obtained:

[0057]

[0058] This system of ordinary differential equations can then describe the motion of the diaphragm 110 (ie, the diaphragm 110 of the transducer 102 and the passive radiator 104) given the driving force from the voice coil 112. However, this does not yet take into account nonlinear behavior.

[0059] Due to the shape of the magnetic field near the voice coil 112, BL is a nonlinear function of the position X1 of the diaphragm 110 of the loudspeaker 102. There are several ways to model this, but a simple approach uses an nth-order polynomial. For example, the following equation can express BL as a function of the position normalized to the rest position multiplied by the nominal value at the rest position:

[0060] BL=(cBL4·x 4 +cBL3·x 3 +cBL2·x 2 +cBL1·x+1)·BL (0) Equation (7)

[0061] While equation (7) shows a 4th order polynomial, it has been recognized that an nth order polynomial can be implemented for equation (7). Due to the physical properties of the suspension of the diaphragm 110, Kms and Rms are nonlinear functions of the position X1. As with BL, Rms and Kms can be expressed as polynomials. The polynomial has been decomposed into two parts, such as a normalized part and a scalar part corresponding to the rest position when X1=0. The benefits of this will become clear in the following refinement:

[0062] Kms=(cK4·x 4+cK3·x 3 +cK2·x 2 +cK1·x+1)·Kms(0) Equation (8)

[0063] Rms=(cR4·x 4 +cR3·x 3 +cR2·x 2 +cR1·x+1)·Rms(0) Equation (9)

[0064] Can be obtained from Figure 5 Equations (8) and (9) are shown in terms of the signal flow through the first normalization circuit 130, the second normalization circuit 132, the first multiplier circuit 134, and the second multiplier circuit 136. Recognizing that cR4.x as depicted in the brackets of equations (8) and (9) 4 Etc. correspond to the first normalization circuit 130 and the second normalization circuit 132, respectively. Each of the first normalization circuit 130 and the second normalization circuit 132 generally includes hardware and software to perform the calculations required by equations (8) and (9).

[0065] In the case of Rms, it can also be a function of the velocity of the diaphragm 110, which can also be modeled as, for example, a polynomial:

[0066] Rms=(cV2·speed 2 +cV1·speed+1)·Rms(x) Equation (10)

[0067] In equation (10), Rms(x) represents Rms of equation (9)

[0068] These equations can then be solved using a numerical method, such as Euler's method, where the equations are iterated at small time steps (small relative to the rate of change of position of any variable in the system 100). In particular, solving the set of equations 1-10 will provide the velocity of the diaphragm 110. This will be described in more detail below.

[0069] Correction by current source

[0070] Now that a model has been developed to estimate the position and velocity of the transducer 102 and the diaphragm 110 of the passive radiator 104, these aspects can be inserted into the system (or audio amplifier system) 150 to correct for distortion (see Figure 6). System 150 can be implemented as a current source amplifier (or audio amplifier) ​​and generally includes an equalization block 152, a core correction block 154, and a transducer prediction model block 156. The computationally simplest approach is to use a current source 158 to drive the voice coil 112. Because of the nature of current source 158, system 150 eliminates the effects of resistance and inductance in voice coil 112 on the current, and these can therefore be canceled. By definition, current source 158 delivers the desired current regardless of the load. In this approach, only the correction current for voice coil 112 may need to be determined.

[0071] The equalization block 152 generates a current target value (or I_target) corresponding to a desired current based on the audio input signal. The transducer model block 160 is typically fed an input current I_vc (or I_correction), which represents the current in the voice coil 112 and is generated by the amplifier 150 in response to at least the target current (i.e., I_target). The transducer prediction model block 156 comprises a combination of hardware and software and calculates the position X1 of the diaphragm 110 of the loudspeaker 102 (or the predicted position of the voice coil 112) according to Equations 2, 3, 6, 7, 8, 9, and 10. The system 150 applies I_correction to the voice coil 112 to move the voice coil 112 to the predicted position of X1 as determined by the transducer prediction model block 156. The transducer prediction model block 156 includes a transducer model block 160, a pressure model block 162, and a passive radiator model block 164. The transducer model block 160 implements Equations 2, 7, 8, 9, and 10. The stress model block 162 generally implements Equation 6, and the passive radiator model block 164 generally implements Equation 3. Given Kms_TD(X1), BL(x), and a target current (I_target from the equalization block 152) from their respective polynomials, a correction current (e.g., I_current) that compensates for nonlinearities in Kms_TD(x) and BL(x) can be calculated as follows:

[0072]

[0073] In general, if BL(x) is less than BL(0), the target current can be increased proportionally, with the increase offsetting the force error due to the change in spring stiffness. However, in such a system, the frequency response may be incorrect because the electrical damping provided by the resistance of the voice coil 112 may be cancelled by the amplifier 150 (or current source). This aspect can be compensated by using a fixed equalization filter in the equalization block 152. Figure 7 express Figure 6 The amplifier 150 also includes a core correction block 155, which can be improved in future implementations.

[0074] Correction by voltage source

[0075] Figure 8 An audio amplifier system 180 is depicted, serving as a voltage source for driving voice coil 112. System 180 includes a current conversion block 182, an adaptation block 184, and a voltage conversion block 186. System 180 provides a correction voltage to voice coil 112 of the transducer in response to an audio input signal. Adaptation block 184 includes a core correction block 190 and a transducer prediction model block 156. Generally, system 180 converts a target voltage (from an equalization block (not shown) (the target voltage is generated based on the audio input signal)) into a target current (i.e., I_target) via current conversion block 182. Core correction block 190 corrects the target current to generate a correction current (i.e., I_correction). Voltage conversion block 186 converts I_correction into a correction voltage (i.e., V_correction) for driving voice coil 112. A voltage source amplifier (not shown) applies V_correction to voice coil 112. System 180 ignores the effects of the inductance of voice coil 112, which is typically effective if the correction is targeted at lower frequencies of system 180. This may be valid because most of the movement and nonlinearity occurs at low frequencies.

[0076] In addition to the position X1 of the diaphragm 110 (see the output from the transducer prediction model block 156), the system 180 also utilizes the predicted velocity of the diaphragm. The current conversion block 182 utilizes the velocity of the diaphragm 110 to convert the audio signal (which is proportional to the voltage) into a target current I_target and transmits the target current I_target to the core correction block 190. The voltage conversion block 186 also converts I_correction into a signal proportional to the voltage to be applied to the voice coil 112. The transducer prediction model block 156 also provides the predicted BL (or the predicted magnetic flux X and the length of the air gap 118). In order to convert I_correction into V_correction according to Equation 13 described below, the voltage conversion block 186 also requires the predicted BL.

[0077] Typically, the target voltage (i.e., the input to the current transformation block 182) must be converted to I_target for use in the transducer prediction model block 156. For example, the movement of the voice coil 112 introduces a current that generates a voltage proportional to the velocity multiplied by "B" multiplied by "L," where "L" corresponds to the length of the air gap; this can be referred to as the back EMF of the voice coil 112. This provides a voltage that is subtracted from the voltage applied to the voice coil 112 (i.e., V_correction) to balance the resistance across the voice coil resistor (e.g., Rvc). The linear target current (i.e., I_correction) that would match the voice coil current if BL(x) were linear can then be calculated using the following equation:

[0078]

[0079] Once the target current is corrected as similarly indicated above, it is necessary to convert the target current back to the correct voltage (i.e., V校正 ). Based on the same relationship, this can be achieved through the following equation:

[0080] V 校正 =1 校正 Rvc 平均 +BL velocity equation (13)

[0081] Changes in voice coil DC resistance (Rvc)

[0082] In a simple approach, it can be assumed that the resistance of the voice coil 112 is constant. Assuming the resistance of the voice coil 112 is constant, Rvc in equation (13) is 平均 Will be set to Rvc 标称 . Typically, the voice coil is formed from copper or aluminum. These materials may experience changes in resistance as their corresponding temperature changes. Therefore, in order to improve the voltage source implementation of the system 180, a thermal model can be used to estimate the temperature rise of the voice coil 112 and, from this, calculate the temperature-corrected resistance of the voice coil 112. Because the current is predicted as I_correct, the power in the voice coil 112 can be obtained. There are several thermal models that can be used based on accuracy. The simplest model can be an RC model, where R represents the thermal resistance of the voice coil 112 to the surrounding environment, and C represents the specific heat capacity of the voice coil 112. The RC model can also be solved iteratively using Euler's method.

[0083] An example of the Euler method for iteratively solving a system of equations is described directly below. By looping through the code of the algorithm shown below over and over again, the algorithm solves the various systems of equations in small time steps so that the equations can be moved within the small time step so as to be considered and viewed as linear. For example, a time step of 200uS (for a sampling rate of 5kHz) can adequately model a typical loudspeaker. Such a model may require downsampling or sampling at the input (e.g., an audio input that can be, for example, 48kHz) and a V-correction and I-correction output that can be 48kHz), and upsampling at the output (e.g., a V-correction and I-correction output that can be 48kHz) using an interpolation filter. Using this approach, a fixed-point full implementation requires approximately 5-6 MIPS / channel for a complete passive radiator system and a minimum of 1-2 MIPS for a closed box system.

[0084] * /

[0085] / / Solve for transducer motion:

[0086] / / dt is defined as the small time step for sampling the system

[0087] X1 = X1 + Velocity_TD*dt (X1 = X1 + velocity TD*dt);

[0088] Force_damping_TD = - Velocity_TD * Rms(X1)_TD (damping force_TD = - velocity TD * Rms(X1)_TD);

[0089] Force_spring_TD=-X1*Kms(X1)_TD

[0090] Force_pressure_TD=−k*pressure*Sd_TD(pressure_TD=−k*pressure*Sd_TD);

[0091] Force_motor = BL(X1)*Ivc_corrected;

[0092] Force_net_TD=Force_damping_TD+Force_spring_TD+Force_pressure_TD+Force_motor (net force_TD=damping force_TD+spring force_TD+pressure_TD+motor force);

[0093] Velocity_TD = Velocity_TD + Force_net_TD / M_TD * dt;

[0094] / / Solve the motion of passive radiation 104:

[0095] Force_damping_PR = -Velocity_PR * Rms(X2,Velocity_PR)PR (damping force_PR = -velocity_PR * Rms(X2,velocity_PR)PR);

[0096] Force_spring_PR=-X2*Kms(X2)_PR(spring force_PR=-X2*Kms(X2)_PR);

[0097] Force_pressure_PR=k*pressure*Sd_PR(pressure_PR=k*pressure*Sd_PR);

[0098] Force_net_PR=Force_damping_PR+Force_spring_PR+Force_pressure_PR (net force_PR=damping force_PR+spring force_PR+pressure_PR);

[0099] Velocity_PR=Velocity_PR+Force_net_PR / M_PR*dt(speed_PR=speed_PR+net force_PR / M_PR*dt);

[0100] X2=X2+Velocity_PR*dt(X2=X2+Velocity_PR*dt);

[0101] / / Solve for the pressure change of shell 101:

[0102] pressure=p_0*(Sd_TD*X1-Sd_PR*X2) / (Vb+Sd*X1+Sd_PR*X2)(pressure=p_0*(Sd_TD*X1-Sd_PR*X2) / (Vb+Sd*X1+Sd_PR*X2));

[0103] / / Calculate the correction current of voice coil 112:

[0104] Ivc_corrected = Ivc_target*BL(0) / BL(X1)+(Kms(X1)-Kms(0)*X1 / BL(X1));

[0105] / / For the voltage source algorithm, add the following C code:

[0106] / / Solve Ivc_target

[0107] Ivc_target = (EQ_out - Velocity_TD * BL(X1)) / Rvoice_coil;

[0108] / / Calculate the correction voltage of voice coil 112:

[0109] V_voicecoil=Ivc_corrected*Rvoice_coil+BL(X1)*Velocity_TD

[0110] Changes in Kms and Rms as a result of movement history

[0111] The model also assumes that Kms and Rms during motion are defined by a polynomial. In reality, these parameters can vary with the "history" of the motion. For example, when the diaphragm 110 moves with significant velocity and displacement, the suspension 114 of the diaphragm 110 may soften. This can change both Rms and Kms.

[0112] As an improvement, estimates of the time-varying values ​​of Rms(0) and Kms(0) can be used to scale the values ​​of Kms and Rms. Since the polynomials for Kms(x) and Rms(x) are normalized to a stationary position, the time-varying parameters can be directly multiplied by the normalized position-varying parameters to determine more accurate Kms and Rms.

[0113] The softening and stiffening of the suspension 114 of the diaphragm 110 as a function of position can be predicted as a time-varying average value that can be modeled as an exponentially decaying sum, where the inputs used for averaging correspond to the steady-state values ​​of Kms and Rms that would result if the amplitude of motion were applied infinitely. This steady-state value of Kms can be expressed as a polynomial equation (14) of the envelope of the varying position.

[0114] Kms 稳态 =a1·|x|+a2 Equation (14)

[0115] Exponential decay can take the form of the following equation.

[0116]

[0117] Then, the average Kms (or Kms) can be calculated by multiplying Equation (15) with Equation (14). 平均 ). This average Kms will then replace Kms(0) in equation (8) to provide:

[0118] Kms=(cK4·x 4 +cK3·x 3 +cK2·x 2 +cK1·x+1)Kms 平均 Equation (16)

[0119] The same form of equation can be used for the Rms steady state

[0120] Rms 稳态 =b1·|x|+b2 Equation (17) Steady-state Rms

[0121] As with Kms, steady-state Rms can be related to the amplitude of motion using equations (15) and (17). The average Rms can then be calculated by multiplying equations (15) and (17). This average Rms will then replace Rms(0) in equation (9) to provide:

[0122] Rms=(cR4·x 4 +cR3·x 3 +cR2·x 2 +cR1·x+1)·Rms 平均Equation (18)

[0123] Kms as stated in Equations 15 and 16 平均 and Rms 平均 The history of the predicted position of the voice coil 112 is taken into account by averaging X1 over the recent history.

[0124] Combined precision over-excursion compression and limiting, frequency compensation, and non-linearity correction

[0125] It is recognized that the embodiments disclosed herein can provide, but are not limited to, advanced speaker protection using precise over-excursion, frequency compensation, and nonlinearity correction, typically without look-ahead processing that may be suitable for amplifier applications including improved auto-tuning power managers. Current implementations of power managers, such as those used in automotive amplifiers, can be difficult to tune manually and may not take into account aspects of the ever-changing environment (such as process, tolerances, aging, etc.). These aspects can result in "guard bands" in protection that can eliminate usable acoustic output, thereby rendering the system quieter. The embodiments herein can combine precise over-excursion limiting with nonlinearity correction and frequency compensation in a manner that does not require look-ahead processing to avoid transient over-excursions.

[0126] One or more of the embodiments disclosed herein, when combined with the adaptive loudspeaker parameter extraction as described in U.S. application Ser. No. 62 / 955,125, filed Dec. 31, 2019, entitled “SYSTEM AND METHOD FOR ADAPTIVE CONTROL OF ONLINE EXTRACTION OF LOUDSPEAKER PARAMETERS” (“the '125 application,” the contents of which are incorporated herein by reference in their entirety), may provide, among other things, an accurate loudspeaker protection mechanism when compared to conventional power manager devices, such as those used in conjunction with automotive amplifiers. One or more of the embodiments may enable a loudspeaker to be reliably driven harder with less margin, and thereby produce a higher sound. Conversely, one or more of the embodiments may also require less margin, which may provide for a lighter loudspeaker design.

[0127] Furthermore, current power managers that provide protection for automotive speaker designs must be manually tuned. This can be time-consuming for engineers involved in developing transducers, amplifiers, and / or digital signal processors (DSPs). Additionally, these implementations may not be adaptive. Current power managers may not be precise enough and may require look-ahead processing to avoid potentially damaging transient over-excursions. Consequently, this aspect may not provide adequate protection for the often very demanding ANC applications.

[0128] The disclosed system(s) and / or method(s) can accurately limit over-excursion, but can also be combined with corrections for transducer nonlinearities to prevent voice coil overheating. Furthermore, because various acoustic implementations can be implemented in real time, such as ANC without the use of a look-ahead delay, any such limiting of over-excursion should be performed without look-ahead processing. Furthermore, because the disclosed limiters for one or more transducers may need to be pushed closer to their excursion limits without increasing the risk of damage, such limiters can tolerate occasional transient over-excursions. Furthermore, the limiters may need to operate within production tolerances, process variations, product lifespan, and environmental conditions (such as temperature). Therefore, the limiters may need to have the capability to allow for automatic tuning. If automatic tuning parameters are available, the disclosed system(s) and / or method(s) may be capable of automatic tuning.

[0129] The disclosed embodiments can improve the power management capabilities for amplifiers (e.g., automotive amplifiers). Existing power managers may not be able to protect against transient over-excursions without look-ahead processing and without considerable margin. This increases the weight and cost of the transducer and does not require careful, time-consuming manual tuning. Furthermore, existing power managers may require a transducer engineer to manually create data tables for a DSP engineer to set up the power manager, and then ultimately for a system engineer to complete the manual tuning. The aspects disclosed herein, when combined with automated tuning of loudspeaker parameters, can eliminate nearly all of the noted drawbacks, including the risk of error and margin requirements.

[0130] Figure 9 A system 200 for providing advanced speaker protection is depicted according to one embodiment. The system 200 can be implemented in an audio amplifier 201 that includes any number of controllers 203 (hereinafter referred to as "controllers 203"). The controllers 203 can be programmed to execute instructions that perform the following operations performed by the system 200 in addition to the systems 350 and 400 described below. The system 200 generally includes a KMS normalization block 130, a BL model block 133, a transducer prediction model block 152, a transducer model block 164, a pressure model block 162, a passive radiator model block 164, a current transformation block 182, a voltage transformation block 186, a filter 202 (e.g., a high-pass filter 202), a limiter block 204, a filter 206 (e.g., a low-pass filter 206), an envelope detector 208, a gain block 210, a first multiplier circuit 212, a second multiplier circuit 214, a divider circuit 216, a conversion block 218, and an adder circuit 220. Generally, the system 200 can protect the loudspeaker 102 from excessive excursion of the voice coil 112. An input audio signal is provided to the current transformation block 182 and the high-pass filter 202.

[0131] System 200 provides an input audio signal through its high-frequency band (e.g., through high-pass filter 202) and low-frequency band (e.g., through low-pass filter 206) for receipt at adder circuit 220. It is recognized that the input audio signal may be, for example, an ANC-based signal. If the input audio signal is limited to the low-frequency band, the signal present in the high-frequency band may not be distorted. Each of high-pass filter 202 and low-pass filter 206 may, for example, operate as a fourth-order filter with a Q of 0.5 and matched corner frequencies. When the low-pass and high-pass signals are added back together through adder circuit 220, this may result in a flat, undistorted frequency response. The corner frequency may be selected, for example, to be approximately 2 to 3 times the resonance of speaker 102, where the movement of voice coil 112 can be sufficiently reduced, as no restriction may be required.

[0132] The current conversion block 182 receives the input audio signal and converts it into a signal representing the input current using Equation 12, which is described above and further below for reference:

[0133]

[0134] Among them, Rvc 标称 is the room temperature DC resistance of the voice coil 112. BL(0) is the voice coil motor force factor when the voice coil 112 is at rest (X1=0). X1 is the position of the voice coil 112. BL can be set to 0 instead of X as described above, and Rvc is set to room temperature. The transducer prediction model block 156 receives the current from the current transformation block (e.g., I 输入 ) to calculate the desired position X1 of the voice coil. In this case, the transducer prediction model block 156 may specify the nonlinear parameters as constant values, for example, as if the desired position X1 of the voice coil 112 is fixed at a stationary position. This may cause the model to be linear. In this case, the transducer prediction model block 156 may determine a calculation for the undistorted position of the voice coil 112, which may have resulted as if the loudspeaker 102 were linear. As part of this calculation, the velocity dx1 / dt is calculated for use in equation (1) above. As described above, the transducer prediction model block 156 (i.e., the linear transducer model 160) may first use Euler's method or other similar iterative numerical method to solve the following equation to obtain X1 (e.g., see equation 2 above, where BL, Kms, Rms are kept constant and thus equation 2 becomes linear).

[0135] As described above, the linear passive radiator model block 164 determines the position of the passive radiator 104 by solving Equation 3 via the Euler method, which is provided again below for reference.

[0136]

[0137] In this case, BL, Kms, and Rms may remain constant, thereby causing Equation 3 to remain constant.

[0138] The pressure model block 162 may then solve for pressure as described above. Thereafter, the pressure model block 162 may solve for pressure "p" according to Equation 6, which is provided above and is also described below for reference.

[0139]

[0140] As described above, the model employed by the pressure model block 162 can be simplified for vented, closed-box, and infinite-baffle acoustic systems. Once the pressure "p" is determined, the linear transducer model block 160 can determine the position (e.g., X1) of the voice coil 112 of the loudspeaker 102. The transducer prediction model block 156 provides the position of the voice coil 112 to the variable gain block (or gain stage) 210 via the second multiplier circuit 214, the limiter block 204, the low-pass filter 206, the divider circuit 216, and the envelope detector 208. When the envelope of the signal provided by the low-pass filter 206 is above the desired maximum displacement, the second multiplier circuit 214 changes the amplitude of the signal. Before this signal reaches the second multiplier circuit 214, the divider circuit 216 rescales the signal to the input signal X1 to achieve a hard knee in the compressor. The second multiplier circuit 214, in combination with the gain block 210, forms the compressor. Gain block 210 performs the function as described in connection with Equation 19, which compares the envelope signal from envelope detector 208 to a threshold. If the envelope is above the threshold, gain block 210 decreases the gain value.

[0141] If the undistorted position X1 is above a predetermined threshold, gain block 210 may reduce the gain applied to position X1 of voice coil 112. For example, divider circuit 216 rescales X1 to a target value at the same scale as X1, and gain block 210 compares X1 to a desired threshold. Limiter block 204 may be active for only a short period of time while reducing the gain applied to position X1 of voice coil 112. Typically, when the envelope encounters a transient, the gain is reduced and limiter block 204 may no longer be needed. For example, Equation 19, as described directly below, provides a method for gain block 210 to adjust the gain.

[0142]

[0143] in:

[0144] δ 阈值 a<1 attenuates x1 envelope x1.

[0145] Envelope detector 208 determines the envelope of position X1 of voice coil 112. For example, envelope detector 208 converts an alternating current (AC) (bidirectional) signal into a DC (unidirectional or forward-only) signal. Envelope detector 208 can then capture the peaks of this signal. Envelope detector 208 can then smoothly control the gain. If envelope detector 208 is not implemented, the gain will only be reduced at the peaks, essentially reverting the system to a simple limiter that is audible and objectionable. If time delay and envelope smoothing are provided, this gradually reduces the undesirable audible characteristics of limiter block 204 alone. Limiter block 204 provides instantaneous detection, but this causes undesirable audible noise when the audio is turned down, which is not preferred. However, if envelope detector 208 is implemented, the undesirable audible portion is gradually reduced so that the listener does not notice it. Because the maximum input to the peak detector is limited (e.g., the input to the envelope detector 208 is limited), the overshoot of the compressor (or collectively referred to as the gain block 210 and the second multiplier block 214) is reduced. However, if this is done, the input needs to be multiplied by 1 / gain first, otherwise the compressor (e.g., the gain block 210 and the second multiplier block 214) will have a limited effect. The divider circuit 216 is provided to provide a hard knee. Without the divider circuit 216, the only way to reduce the gain is if the target position X1 of the voice coil 112 is increased, which results in a soft knee and therefore poor control. For example, the volume increases indefinitely (e.g., a soft knee scenario). With the divider circuit 216, if the volume gradually increases until it reaches the expected maximum value that cannot be exceeded, a hard knee characteristic is exhibited.

[0146] Additionally or alternatively, the input to the peak detector can be taken before the gain multiplication stage (not shown). In this case, the input may not need to be multiplied by 1 / gain. However, preventing the gain block 210 from having any overshoot may require a slower attack rate, which will force the limiter block 204 to be more aggressive and more audible. In all cases, the attack rate of the envelope detector 208 can be optimized to prevent the gain block 210 from over-compressing. This can be in the range of, for example, tens of milliseconds. In addition, the envelope detector 208 can have a slow release to prevent the gain block 210 from pumping or releasing and attacking with every peak or transient. The release time can be, for example, on the order of several hundred milliseconds.

[0147] Once the gain block 210 (and second multiplier block 214) compress the output of the envelope detector 208, the limiter block 204 can limit the excursion and temperature of the voice coil 112. Generally, once the position signal (e.g., the position X1 of the voice coil 112) has been compressed by the gain block 210, the limiter block 204 can then limit the signal. For example, once the undistorted position signal has passed through the gain multiplication stages (e.g., gain block 210, second multiplier circuit 214, and divider circuit 216), it can then be presented to the limiter block 204. The limiter block 204 can then limit the positive and negative positions to at least one predetermined maximum value that can be safe for the transducer 102. The limiter block 204 typically accounts for sudden and high-level transients that may not be adequately compressed due to attack delay. If this were allowed to occur, the voice coil 112 could strike a backplate (not shown) positioned above the transducer 102 and be damaged.

[0148] First multiplier circuit 212 can multiply the output of gain block 210 by the audio output of high-pass filter 202. This aspect can maintain a roughly equal balance between high and low frequencies, which can be less objectionable than simply reducing the low frequencies. Once signal X1 is compressed and limited by gain block 210 and limiter block 204, respectively, the signals can be provided as X1_target to pressure model block 162 and passive radiator model block 164 (e.g., see secondary model block 230). Secondary model block 230 can determine the velocity, pressure, and nonlinear parameters of diaphragm 110. Because the nonlinear elements of transducer 102 can be corrected, the resulting position of voice coil 112 in the next stage of the process can be the same as the undistorted position of voice coil 112. Pressure model block 162 can calculate the pressure in housing 101 using Equation 4, and passive radiator model block 164 can calculate the position of the passive radiator using Equation 5. For example, Equations 4 and 5 can be solved again using Euler's method or other suitable comparable numerical techniques.

[0149] Typically, it may be necessary to convert back to the desired current based on Equation 20 as described below. For example, the conversion block 218 may convert the output from the low-pass filter 206, the secondary model block 230, the KMS normalization block 130, and the BL model block 133 into a target current (I 目标 Since Equation 6 uses the nonlinear parameters as described above to correct nonlinear distortion, the following equation is used to calculate the desired voice coil current (ie, the target current (I 目标 )).

[0150]

[0151] Generally, Equation 20 sets the way in which nonlinear parameters can be solved. Since all inputs to the transformation block 218 are known (or Equation 20 is known), the transformation block 218 may need to obtain the derivative and the 2nd derivative of X1_target and solve for the target current I 目标 However, as shown in Equation 20, for the conversion block 218 for correcting the nonlinear elements KmsTD and BL; such elements can be nonlinear KmsTD(x1_ 目标 ) and BL(x1_ 目标 ). These values ​​can be calculated as described above and can be further provided in conjunction with Equation 7 and Equation 8 directly below for reference.

[0152] BL=(cBL4·x 4 +cBL3·x 3 +cBL2·x 2 +cBL1·x+1)·BL(0) Equation (7)

[0153] as well as

[0154] Kms=(cK4·x 4 +cK3·x 3 +cK2·x 2 +cK1·x+1)·Kms 平均 Equation (8)

[0155] Furthermore, if the Kms average and RmsTD are periodically updated from a real-time system that extracts these parameters, the system 200 can be made tunable to automatically tune and compensate for frequency changes. Aspects of providing extraction techniques, such as utilizing a bandpass filter, will be described in greater detail below. Generally, one or more embodiments can provide a blend of correction for nonlinear distortion and position limiters by providing an appropriate predistortion voltage to the voice coil 112.

[0156] If the amplifier 201 is configured as a current source, the target current I 目标 Since most amplifiers are configured as voltage sources, I 目标 Can be converted into voltage. For example, the voltage conversion block 186 can use the following equation to convert the target current I 目标 Convert to voltage target value V 目标 :

[0157]

[0158] If a nonlinear parameter of BL(x) is used, the correction can be performed using Equation 21. The adder circuit 220 adds the output of the high-pass filter 202 (eg, the high-frequency input audio signal) to the voltage target value V 目标 The voltage target value V目标 Typically corresponds to an amount of voltage to drive / move the voice coil 112 to a desired position without experiencing excessive excursion and over-temperature conditions.

[0159] It is recognized that nonlinear elements can be ignored and therefore not utilized in Equations 7 and 8. However, if Approaches 7 and 8 are not used, errors may exist. For example, this may result from the assumption that X1_target and X1 are no longer valid in a real speaker. However, if the primary goal is to protect speaker 102, such errors may be small enough to be ignored.

[0160] In addition, it is possible to eliminate the high-pass / low-pass filter structure (e.g., high-pass filter 202 and low-pass filter 206). Although the performance of system 200 may be degraded, this degradation may be acceptable in some cases. For example, the elimination of the high-pass / low-pass structure may degrade the incoming audio signal due to increased distortion from limiter block 204 and because limiting low-frequency signals may also distort coexisting high-frequency signals. It is also possible to include some of the other model elements described above to improve the model, particularly in cases where the Kms mean and Rms mean are not extracted separately.

[0161] Figures 10 to 12 Graphs 250, 252, and 254 are generally provided illustrating the behavior of the compressor (or gain block 210) and limiter block 204, and the speaker 102, respectively, according to one embodiment. For example, Figures 10 to 12 The following diagram generally illustrates the behavior of gain block 210 and limiter block 204, along with an actual loudspeaker, when a sudden, large signal is applied and removed. Waveform 260 corresponds to the position of voice coil 120 as it moves in and out during a high-power transient. Waveform 262 corresponds to the gain of gain block 210 when the compressor engages to reduce the excessive signal. As can be seen, the delay in the compressor gain reduction allows the initial excessive excursion that could damage loudspeaker 112 to occur. Figure 11 A slow attack is generally shown to avoid over-compression and allow for a large amount of over-excursion and primarily low frequency artifacts for the voice coil 112. In this case, there may not be over-compression, however many transients may be present (e.g., for automotive applications such as road noise cancellation, this could be a stray drum, bass pluck, or bump in the road).

[0162] Figure 12 A fast attack is shown which avoids low frequency artifacts while still allowing for voice coil 112 excursion. Figure 12The waveform 260 depicts the expected maximum excursion of the voice coil 112. Excessive compression can cause the compressor (or gain block 210) to pump on every transient, which can be annoying to the listener. In other words, if the attack of the gain block 210 is too fast, the gain block 210 over-compresses, which causes the audio or sound to be muffled, as if the volume is being modulated. With this embodiment, a limiter block 204 that provides a slower attack and faster release can be utilized, which does not pump the gain block 210 (or even briefly over-excursions). This is perceived as inaudible, which may be the intention.

[0163] Figure 13 A graph 256 is provided that depicts the effect of the limiter block 204 controlling the maximum position without the use of a compressor (or gain block 210). Graph 256 shows that the limiter block 204 controls the maximum position without the compressor 210. In effect, this illustrates the position being truncated by control to avoid damaging the voice coil 112 of the transducer 102. In general, graph 256 illustrates the behavior or if the limiter block 204 were active on its own (e.g., the envelope detector 208, gain block 210, and second multiplication circuit 214 are engaged to reduce the gain) without the compressor. Graph 256 further illustrates that the displacement of the voice coil 112 is limited to the desired maximum displacement.

[0164] Extraction technique (using bandpass filter)

[0165] As previously described, the system 200 can be made to automatically tune or adapt to changing parameters of the speaker 102. For example, the eight tracking bandpass filters can be grouped into four sets of two filters. One set of filters can track the maximum impedance that exists at the resonant frequency. The second set of filters can track the minimum impedance that exists above the resonant frequency of the speaker 102. The third and fourth sets of filters can track the -3dB points in the impedance curve above and below the resonant frequency of the speaker 102, where the impedance is half of the maximum impedance. For each set of two filters, the inputs can be the voice coil voltage and current. The output of each filter can be converted to an RMS (root mean square) value. Then, at each set of filter bandpass frequencies, the impedance is the RMS value corresponding to the voltage divided by the RMS value corresponding to the current. Once these values ​​are known, the Q of the system (e.g., the quality of a mechanical system (Q ms ), the quality of the electrical system (Q ES ) and the quality of the total (complete) system (Q TS) can be calculated from the half impedance point by definition. Generally, the quality factor Q is a defined engineering term and, for a loudspeaker, may be related to the bandwidth of a resonant peak in the impedance frequency response. The resonant frequency may be the frequency of a bandpass filter that tracks the impedance maximum. The minimum impedance can be used as a good approximation of the DC resistance of the voice coil 112. Based on Q, F 共振 and Rdc; The average Kms and Rms can be calculated for a closed box or infinite baffle acoustic system based on the following relationship.

[0166] The following disclosure provides how Q, F resonance, and Rdc are related to Kms(avg) (Eq. 23) and Rms(avg) (Eq. 13) and Mms (see Eq. 12 below).

[0167] Based on the maximum impedance Zmax and Rdc, the following values ​​can be calculated:

[0168]

[0169] Based on the result of equation 22 directly above, it is possible to calculate 1 / 2Pi x F 共振 and Qts to determine the average Kms:

[0170]

[0171] Based on the result of Equation 22, calculate 1 / 2Pi x F 共振 To determine the following values:

[0172] M MS =τ T 2 ·K MS Equation (24)

[0173] Determine the average Rms based on Zmax and Rdc:

[0174]

[0175] If BL is unknown, a normalized value of 1 can be used. However, this aspect may require matching the threshold of the displacement limit to be calibrated. For example, by measuring a sudden increase in distortion in voice coil 112, the current can be increased as the magnitude of the displacement. This aspect can then correspond to the limiter threshold and be used to scale the calculated normalized displacement to the correct level. If BL is unknown, it is possible to at least calibrate the point where the displacement is too high, which can be detected by a sudden increase in distortion in the voice coil current. The distortion fingerprint from the '125 application can be used for the maximum displacement.

[0176] Alternatively, the above system of equations can be solved instead, where Mms is known or normalized to 1, and Kms, BL, and Rms are solved for. Because the tracking bandpass filter output has a noise floor below a certain minimum signal level in any one band, the output may not be usable. To prevent the system from becoming unstable under these conditions, the last known good values ​​of the average Kms and average Rms are used until new good values ​​are obtained. Often, there are signals for which a BP filter implementation may not be possible, but these signals can be slowed down. There are several implementations that can achieve tracking. One embodiment may include using feedback to adjust the tracking frequency up or down based on whether the impedance is decreasing or increasing.

[0177] Online adaptive parameter extraction

[0178] The '125 application, as described above, introduces the concept of multiple parameters associated with loudspeaker 102 into the realm of online and adaptive audio. For example, the '125 application describes one or more audio systems that can provide the resistance of voice coil 112 (e.g., Rdc), the estimated resonant frequency of loudspeaker 102 (e.g., fres), the resistance of loudspeaker 102 at the resonant frequency (e.g., Res), the overall (complete) system quality (e.g., Qts), the impedance of loudspeaker 102, etc. These characteristics may be based on, among other things, determining the admittance curve of loudspeaker 102. By having instant access to these parameters, it is possible to control, among other things, the maximum excursion of voice coil 112 and provide a thermal limiter to protect loudspeaker 102 from damage, as described below.

[0179] Overtemperature protection

[0180] Figure 14 A system 350 for protecting a loudspeaker 102 from an over-temperature condition of a voice coil 112 is depicted according to one embodiment. Generally, the system 350 includes the above-described Figure 9 Part of the system 200 described (e.g., in terms of over-excursion protection provided by the system 200), and preceding it, is a thermal protection mechanism that can reduce the level of the input audio signal when the temperature of the voice coil 112 is above a predetermined temperature threshold that can potentially damage the voice coil 112.

[0181] System 350 includes a power calculation block 352, a thermal model block 354, an average calculation block 356, a rated power block 358, a comparator circuit 360, a unification block 361, a calculation reduction block 362, a multiplier block 364, and an offset protection block 366. System 350 also includes a detector block 208, a gain block 210 (or compressor 210), a first multiplication block 212, and a divider circuit 216. Power calculation block 352 determines the power loss in voice coil 112 by first disconnecting, determining voice coil current Ivc, squaring voice coil current Ivc, and then dividing the squared value of Ivc by the DC resistance Rdc of voice coil 112. It is recognized that Rdc can be obtained through the disclosure of the '125 application, and that utilizing Rdc through the '125 application can provide improved accuracy.

[0182] If obtaining Rdc using the methods described in the '125 application is not possible, the resistance Rdc of the voice coil 112 can be calculated by taking the temperature rise and thermal coefficient of resistance of the voice coil 112. Typically, the resistance Rdc and the amount of change in Rdc are known. Therefore, the temperature can be derived from this. For example, because the metal in the voice coil wire changes its resistance with temperature, the temperature can be calculated from the known resistance. If direct measurements are not available, the thermal model block 354 can determine the temperature. For example, the thermal model block 354 can determine the temperature after receiving the power loss in the voice coil 112 from the power calculation block 352. The thermal model block 354 can employ a simple first-order thermal model that utilizes the thermal resistance between the voice coil 112 and the surrounding environment and the thermal capacity of the voice coil 112, both of which are modeled in parallel with the voice coil power loss as current.

[0183] The voice coil current may be measured using appropriate hardware such as, for example, a current sense and an analog-to-digital (A-to-D) converter (both not shown). However, if this hardware is not available in system 350, the voice coil current may be measured from Figure 9 The transducer prediction model block 152 obtains the current. The thermal model block 354 can then provide the temperature of the voice coil 112 to the gain block 210 (e.g., via the divider circuit 216 and envelope detector block 208 as described above). In this case, the attack and release speeds can be measured in seconds rather than milliseconds. The attack and release can be in a time frame similar to the thermal time constant of the system. If the attack and release are too fast, the compressor (e.g., envelope detector 208, gain block 210, and second multiplier circuit 214) may overreact. Conversely, if the attack and release are too slow, the compressors 208, 210, and 214 may underreact.

[0184] Furthermore, since the power loss is known (e.g., as calculated by power calculation block 352), average value calculation block 356 receives the power loss of voice coil 112 and determines the average power of the power loss. Comparator 360 determines whether the average power, as the output from average value calculation block 356, is greater than the rated power provided by rated power block 358. If the average power is less than the rated power, comparator 360 provides the output to unification block 361, which multiplies the output by 1. Therefore, no gain change occurs, and the output of unification block 361 is then provided to multiplier block 364.

[0185] However, if the average power is greater than the rated power, comparator 360 provides its output to square root block 362. In turn, computational reduction block 362 reduces the signal level by the square root of the rated power divided by the average power. Because power is proportional to the square of the signal level, computational reduction block 362 can utilize the square root. The average power can be estimated over a longer period of time, similar to the thermal time constant of voice coil 112. Measured or calculated power loss can be used, and the temperature model block 354 can then be used to determine the temperature. Generally, multiplier circuit 364 and / or divider circuit 362 can adjust the amplitude of signal Vtarget provided to loudspeaker 102.

[0186] Because the average power is too high (e.g., higher than the rated power), the excursion protection block 366 is used to reduce the incoming signal Vtarget, then the excursion of the voice coil will be smaller, but since this protection is related to the average value, excursion protection may still be needed when the transient value may be much higher than the average value. Typically, the excursion protection block 366 is performed in conjunction with Figure 9 The offset protection block 366 generally includes the KMS normalization block 130 , the BL model block 133 , the voltage transformation block 186 , the limiter block 204 , the low-pass filter 206 , and the conversion block 218 .

[0187] Figure 15A system 400 for providing accurate measurement of the indirectly measured temperature of the voice coil 112, according to one embodiment, is depicted. This approach uses the same bandpass filter concept as described above (e.g., the minimum frequency where the impedance is minimum (see also the '125 application)). For example, the system 400 includes bandpass filters 402, 404, absolute value blocks 406, 408, average calculation blocks 410, 412, a divider circuit 414, and a temperature calculation block 416. Each of the bandpass filters 402, 404 can have a narrow passband frequency that is tuned to a location near the minimum impedance of the voice coil 112 above the resonance of the loudspeaker 102. Thus, the bandpass filter 402 allows frequencies on the voltage output of the voice coil 112 (e.g., Vvc) to pass to the absolute value block 406, corresponding to the minimum impedance of the voice coil 112 above the resonance of the loudspeaker 102. Similarly, the bandpass filter 404 enables frequencies on the current output from the voice coil 112 (eg, Ivc) that correspond to the minimum impedance of the voice coil 112 above which resonance of the loudspeaker 102 occurs to pass to the absolute value block 408 .

[0188] The divider circuit 414 divides the average of the absolute values ​​of the voltage Vvc by the average of the absolute values ​​of the current Ivc to provide the magnitude of the impedance at the minimum impedance (e.g., to provide the resistance Rdc of the voice coil 112). This impedance may be determined by the Rdc of the voice coil 112. Therefore, a first approximation may be taken as the magnitude of Rdc. Once Rdc is known, and then by using the thermal coefficient of resistance of the voice coil 112, the temperature calculation block 416 may determine the temperature. The temperature may be used in place of the previously mentioned value from the thermal model block 354 (see Figure 14 ) because the temperature determined by the temperature calculation block 416 may be more accurate. However, this approach requires measuring the current through the voice coil.

[0189] In general, the above method is sufficient if there is sufficient signal energy at the frequencies of the bandpass filters 402, 404. If not, the results may be erroneous and are preferably ignored. This can be achieved by comparing the average of the absolute values ​​of the currents to a threshold. If the average of the absolute values ​​of the currents is below the threshold where noise may become a problem, the results should be ignored. If this is the case, then instead use Figure 14 The modeling temperature described in .

[0190] Figure 16 A method 500 for providing advanced speaker protection is depicted according to one embodiment. In operation 502, the audio amplifier 201 receives an audio input signal.

[0191] In operation 504, the transducer prediction model block 156 generates an excursion signal X1 corresponding to a first excursion level of the voice coil 112 based on the audio input signal. Figure 9 As shown, the transducer prediction model block 156 utilizes, among other things, the pressure in the housing 101 associated with the speaker 102 to generate an excursion signal X1.

[0192] In operation 506, the limiter block 204 limits the excursion signal X1 to achieve a maximum excursion level X1_target. For example, the limiter block 204 generates the maximum excursion level X1_target. In operation 508, the secondary model block 230 determines a target pressure (P_target) of the housing 101 associated with the speaker 102 based on the maximum excursion level X1_target. In operation 510, the conversion block 218 generates a target current signal (i_target) based on at least the target pressure (P_target) of the housing 101. 目标 In operation 512, the voltage conversion block 186 converts the target current signal (i 目标 ) is converted into a target voltage signal (v 目标 ) (or drive signal) to drive the voice coil 112 to a maximum excursion level (eg, X1_target).

[0193] Although exemplary embodiments have been described above, it is not intended that these embodiments describe all possible forms of the invention. Rather, the words used in this specification are words of description rather than limitation, and it should be understood that various changes can be made without departing from the spirit and scope of the invention. In addition, the features of the various implemented embodiments can be combined to form other embodiments of the invention.

Claims

1. An audio amplifier system, comprising: a loudspeaker comprising a voice coil for generating an audio output into a listening environment; as well as an audio amplifier operatively coupled to the speaker and programmed to: receiving an audio input signal; generating an excursion signal corresponding to a first excursion level of the voice coil based on the audio input signal; limiting the excursion signal to a maximum excursion level; and determining a target pressure for a housing of the loudspeaker based on the maximum excursion level; generating a target current signal based at least on the target pressure; and The target current signal is converted into a target voltage signal to a target driving signal to drive the voice coil to reach the maximum excursion level. 2 . The audio amplifier system of claim 1 , wherein the audio amplifier is further programmed to apply a first filter to the maximum excursion level before determining the target pressure of the housing. The audio amplifier of claim 2 , wherein the first filter is a low-pass filter.

4. The audio amplifier system of claim 1 , wherein the audio amplifier includes a compressor programmed to compress the excursion signal before limiting the excursion signal to the maximum excursion level. 5 . The audio amplifier system of claim 1 , wherein the audio amplifier includes a compressor programmed to receive a maximum excursion limit before determining the target pressure to control a gain of the maximum excursion limit. 6 . The audio amplifier of claim 1 , wherein the audio amplifier is further programmed to generate the target current signal based on a stiffness of a diaphragm of the speaker.

7. The audio amplifier of claim 1, wherein the audio amplifier is further programmed to apply a first filter to the audio input signal.

8. The audio amplifier of claim 7, wherein the first filter is a high-pass filter.

9. The audio amplifier of claim 7, wherein the audio amplifier is configured to apply the target voltage to the output of the first filter before driving the voice coil to the maximum excursion level.

10. A computer program product embodied in a non-transitory computer-readable medium, the computer program product being programmed to protect a speaker, the computer program product comprising instructions for: receiving an audio input signal; generating an excursion signal corresponding to a first excursion level of a voice coil of the loudspeaker based on the audio input signal; limiting the excursion signal to achieve a maximum excursion level; and determining a target pressure for a housing of the loudspeaker based on the maximum excursion level; generating a target current signal based at least on the target pressure; and The target current signal is converted into a target voltage signal to a target driving signal to drive the voice coil to reach the maximum excursion level.

11. The computer program product of claim 10, further comprising: A first filter is applied to the maximum excursion level before determining the target pressure for the housing.

12. The computer program product of claim 11, wherein the first filter is a low pass filter.

13. The computer program product of claim 10, further comprising: The excursion signal is compressed before being limited to achieve the maximum excursion level.

14. The computer program product of claim 10, further comprising: Before determining the target pressure, a maximum excursion limit is received to control a gain of the maximum excursion limit.

15. The computer program product of claim 10, further comprising: The target current signal is generated based on a hardness of a diaphragm of the speaker.

16. The computer program product of claim 10, further comprising: A first filter is applied to the audio input signal.

17. The computer program product of claim 16, wherein the first filter is a high pass filter.

18. The computer program product of claim 16, further comprising: The target voltage is applied to the output of the first filter before driving the voice coil to the maximum excursion level.

19. A method for protecting a loudspeaker, the method comprising: receiving an audio input signal; generating an excursion signal corresponding to a first excursion level of a voice coil of the loudspeaker based on the audio input signal; limiting the excursion signal to achieve a maximum excursion level; determining a target pressure for a housing of the loudspeaker based on the maximum excursion level; generating a target current signal based at least on the target pressure; as well as The target current signal is converted into a target voltage signal to a target driving signal to drive the voice coil to reach the maximum excursion level.

20. The method of claim 19, further comprising: A first filter is applied to the maximum excursion level before determining the target pressure for the housing.

Images