Power amplifier co-linearization method and system

Abstract

The application is suitable for the field of amplifier technology, and provides a power amplifier cooperative linearization method and system, the method comprising: obtaining an error signal of a power amplifier, and separating the error signal into slow-varying and fast-varying error components; performing predictive feedforward control based on the fast-varying component to suppress the short-time electrical memory effect of the power amplifier; and performing decoupling compensation control based on the slow-varying component to compensate for the long-time thermal memory effect of the power amplifier. The decoupling compensation control orthogonally decomposes the slow-varying error component into in-phase and quadrature components to independently compensate for amplitude and phase distortion. Through fast-slow loop cooperation, predictive intervention and decoupling compensation, the application can quickly respond to transient distortion and accurately compensate for long-term drift, thereby improving the dynamic linearity, compensation accuracy and stability of the power amplifier.

Description

Technical Field

[0001] This application belongs to the field of amplifier technology, and in particular relates to a method and system for cooperative linearization of power amplifiers. Background Technology

[0002] Power amplifiers are core components in wireless communication systems, their main function being to amplify low-power signals to a power level sufficient for communication distance requirements. However, the inherent nonlinear characteristics and memory effects of power amplifiers can cause signal distortion, severely impacting communication quality. Memory effects primarily include short-term electrical memory effects caused by bias network resonance, and long-term thermal memory effects caused by changes in device junction temperature.

[0003] Existing linearization techniques have shortcomings in dealing with complex dynamic conditions. On the one hand, for distortions triggered instantaneously by drastic changes in the input signal envelope, many existing solutions lack predictive analysis of the signal's dynamic characteristics, resulting in passive compensation mechanisms and lag in dynamic response. On the other hand, while existing technologies may include elements such as feedforward control or orthogonal decomposition for signal processing, they lack a unified collaborative control framework. Specifically, existing technologies typically process the total error signal directly, failing to clearly separate the total error signal into fast-changing components corresponding to short-term electrical memory effects and slow-changing components corresponding to long-term thermal memory effects, and failing to apply optimized control with different mechanisms to each. Furthermore, when dealing with distortions caused by slow-changing factors such as thermal effects, existing technologies struggle to effectively decouple and compensate for gain drift, i.e., amplitude distortion and phase drift, leading to mutual interference between amplitude and phase compensation loops, thus limiting the long-term stability and final accuracy of the compensation. Summary of the Invention

[0004] This application provides a power amplifier cooperative linearization method and system, which aims to solve the problems of slow dynamic response, low compensation accuracy and poor long-term stability in existing power amplifier linearization technologies. In particular, it addresses the shortcomings of existing technologies in coordinating and efficiently dealing with fast and slow distortion, failing to upgrade passive response to active prediction, and failing to decouple and compensate for amplitude and phase components in slow distortion.

[0005] In a first aspect, embodiments of this application provide a power amplifier cooperative linearization method, including:

[0006] Obtain the error signal that reflects the output distortion of the power amplifier, and separate the error signal into a slow-changing error component and a fast-changing error component;

[0007] Predictive feedforward control based on the fast-changing error components is performed to suppress the short-time electrical memory effect of the power amplifier;

[0008] Decoupling compensation control based on the slowly varying error component is performed to compensate for the long-term thermal memory effect of the power amplifier.

[0009] In one possible implementation of the first aspect, separating the error signal into a slowly varying error component and a rapidly varying error component includes:

[0010] The slowly varying error component is extracted from the error signal using a low-pass filter;

[0011] The rapidly changing error component is extracted from the error signal using a bandpass filter.

[0012] In one possible implementation of the first aspect, the execution of predictive feedforward control based on the rapidly varying error components includes:

[0013] Acquire input signal;

[0014] The input signal is analyzed to identify power transient events, and composite transient feature parameters are extracted from the power transient events.

[0015] Based on the aforementioned composite transient characteristic parameters, the dynamic gain factor is calculated;

[0016] Based on the rapidly changing error components, a basic cancellation signal is generated;

[0017] The amplitude of the base cancellation signal is modulated using the dynamic gain factor to obtain the final cancellation signal;

[0018] The final cancellation signal is injected into a bias line of the power amplifier.

[0019] In one possible implementation of the first aspect, the composite transient characteristic parameters include transient kurtosis and transient peak power;

[0020] The calculation of the dynamic gain factor based on the composite transient characteristic parameters includes:

[0021] The dynamic gain factor is calculated using the following formula:

[0022] γ=1+k_slew*S_transient+k_peak*(P_peak_transient-P_avg);

[0023] Wherein, γ is the dynamic gain factor; S_transient is the transient steepness; P_peak_transient is the transient peak power; k_slew is the preset steepness gain coefficient; k_peak is the preset peak power gain coefficient; and P_avg is the average input power within a preset time period.

[0024] In one possible implementation of the first aspect, the bias line includes a gate bias line and a drain bias line;

[0025] The process includes, before injecting the final cancellation signal into a bias line of the power amplifier:

[0026] Spectral analysis is performed on the signals on the gate bias line and drain bias line of the power amplifier to determine the distribution of resonant energy;

[0027] The step of injecting the final cancellation signal into a bias line of the power amplifier includes: injecting the final cancellation signal into a bias line where the resonant energy is dominant, based on the distribution of the resonant energy.

[0028] In one possible implementation of the first aspect, the execution of decoupling compensation control based on the slowly varying error component includes:

[0029] The slowly varying error component is orthogonally decomposed into an in-phase error component reflecting gain drift and an orthogonal error component reflecting phase drift;

[0030] The amplitude distortion of the power amplifier is compensated based on the in-phase error component.

[0031] The phase distortion of the power amplifier is compensated based on the orthogonal error components.

[0032] In one possible implementation of the first aspect,

[0033] The compensation for amplitude distortion of the power amplifier based on the in-phase error component includes:

[0034] An adaptive temperature compensation model is used to determine the compensation current and perform amplitude compensation. Based on the cumulative value of the in-phase error component, a least squares optimization algorithm is used to iteratively optimize at least one temperature compensation coefficient in the adaptive temperature compensation model.

[0035] The compensation for phase distortion of the power amplifier based on the orthogonal error components includes:

[0036] An adaptive phase correction model is used to determine the pre-corrected phase shift and perform phase compensation; based on the cumulative value of the orthogonal error components, a least squares optimization algorithm is used to iteratively optimize at least one phase correction coefficient in the adaptive phase correction model.

[0037] In one possible implementation of the first aspect, the method further includes:

[0038] Monitor the long-term drift trends of the temperature compensation coefficient in the adaptive temperature compensation model and the phase correction coefficient in the adaptive phase correction model;

[0039] Based on the long-term drift trend, predictive maintenance information about the health status of the power amplifier is generated.

[0040] Secondly, embodiments of this application provide a power amplifier cooperative linearization system, comprising:

[0041] An error processing unit is used to acquire an error signal and separate the error signal into a slowly varying error component and a rapidly varying error component.

[0042] A feedforward compensation unit is used to generate a cancellation signal and inject the cancellation signal into the bias circuit of the power amplifier;

[0043] The first compensation unit is connected to a power stage of the power amplifier and is used to compensate for the amplitude distortion of the power amplifier.

[0044] A phase pre-correction module is connected in series in one of the input signal paths of the power amplifier to compensate for the phase distortion of the power amplifier.

[0045] The control unit, connected to the error processing unit, is used to acquire the input signal of the power amplifier, receive the slow-varying error component and the fast-varying error component, and is used to: control the feedforward compensation unit to perform predictive feedforward control based on the fast-varying error component; and control the first compensation unit and the phase pre-correction module to perform decoupling compensation control based on the slow-varying error component.

[0046] This application's embodiments, by introducing predictive feedforward control based on the dynamic characteristics of the input signal, transform passive feedback into active intervention. This allows for the pre-amplification of the suppression signal before distortion occurs, shortening the system response time and improving the dynamic linearity of the power amplifier when processing peak-to-average power ratio (PAPR) signals. Furthermore, by orthogonally decomposing the slow-varying error into independent sub-components representing amplitude and phase, and guiding gain and phase drift compensation respectively, the problem of amplitude and phase coupling in traditional slow-loop compensation is alleviated, resulting in a substantial improvement in the accuracy and stability of long-term compensation. Furthermore, a closed-loop control system with fast and slow loops working in tandem is constructed, capable of simultaneously and efficiently handling nonlinear problems at different time scales and physical dimensions, resulting in a more comprehensive and thorough compensation effect. Finally, by monitoring the long-term drift trend of key coefficients in the adaptive compensation model, the aging state of devices can be accurately predicted and predictive maintenance information can be reported, realizing a shift from fault repair to predictive maintenance and improving the long-term reliability and maintainability of the system. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A schematic diagram of a power amplifier cooperative linearization system provided in this application embodiment;

[0049] Figure 2 A flowchart illustrating a power amplifier cooperative linearization method provided in this application embodiment;

[0050] Figure 3 A schematic diagram illustrating the principle of error signal separation, decomposition, and allocation provided in an embodiment of this application;

[0051] Figure 4 A schematic diagram of a slow loop decoupling compensation adaptive model provided in an embodiment of this application;

[0052] Figure 5 An interaction timing diagram of key signaling under a hybrid control architecture provided in this application embodiment;

[0053] The accompanying drawings are described below:

[0054] 10-Power amplifier; 20-Phase pre-correction module; 30-First compensation unit; 40-Second compensation unit; 50-Error processing unit; 60-Control unit; 70-Adaptive temperature compensation model; 80-Adaptive phase correction model. Detailed Implementation

[0055] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0056] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0057] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0058] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0059] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0060] References to "one embodiment" or "some embodiments" in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Those skilled in the art should understand that the technical solutions of this application can have various modifications and equivalent substitutions. For example, the specific filter types, amplifier categories, parameter values, or digital implementation platforms (such as FPGAs or DSPs) described in the embodiments are exemplary and can be adjusted according to specific performance requirements, cost budgets, and technical conditions. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0062] Example 1

[0063] This application provides a power amplifier cooperative linearization system and its corresponding operating method. This scheme innovatively integrates multiple techniques such as error separation, predictive feedforward control, and decoupling compensation control to form a cooperative closed-loop control system, addressing the complex nonlinear distortion problem faced by power amplifiers in modern communication systems.

[0064] Reference Figure 1 The figure is a schematic diagram of the structure of a power amplifier cooperative linearization system provided in an embodiment of this application. Figure 1 As shown, the system is built around a core device—a power amplifier 10—that needs to be linearized. The power amplifier 10 can be any power amplifier requiring linearization, particularly suitable for applications such as modern wireless communication base stations, satellite communication terminals, or broadcast television transmitters. For example, the power amplifier 10 can be a power amplifier based on a laterally diffused metal-oxide-semiconductor (LDMOS) transistor, a gallium arsenide (GaAs) transistor, or a gallium nitride (GaN) transistor. Although these different types of power amplifiers differ in materials and processes, they generally suffer from nonlinear distortion caused by both electrical and thermal memory effects, which is precisely the technical challenge that this application aims to address.

[0065] Specifically, the system's workflow begins with the input signal path. An input signal (e.g., a modulated communication signal) is first sent to the phase pre-correction module 20. This phase pre-correction module 20, connected in series in the main input signal path of the power amplifier 10, is essentially a voltage- or digitally controlled phase shifter. This module applies a precise and dynamically variable phase shift to the input signal according to control commands issued by the subsequent control unit 60. Its main purpose is to pre-compensate for phase distortion in the power amplifier 10 caused by slowly varying factors such as thermal effects, thereby canceling out some phase drift at its source by pre-correcting the signal before it enters the amplifier.

[0066] The phase-precorrected signal then enters power amplifier 10 for power amplification to achieve the required transmission power of the system. At the output of power amplifier 10, a directional coupler or similar signal sampling device is connected. Its function is to couple out a small, proportionally powerful sampled output signal from the high-power output signal without distortion. This sampled output signal is vector-compared with a reference signal from the input (typically the original input signal after appropriate delay and amplitude adjustment) in a comparator circuit (e.g., vector subtraction via a mixer and subtractor) to generate an error signal. This error signal accurately reflects, in both amplitude and phase, all distortions introduced by power amplifier 10 from input to output; in this application, it may be referred to as the total error signal. This total error signal is the sole basis for calculations and decisions in all subsequent compensation control loops.

[0067] The total error signal is transmitted to the error processing unit 50. The core function of the error processing unit 50 is to separate the complex total error signal in the frequency or time domain to extract distortion components with different time-scale characteristics, thus achieving a divide-and-conquer approach. Specifically, it separates the total error signal into a slowly varying error component and a rapidly varying error component. It is important to note that this separation is based on a deep understanding of the physical mechanisms of distortion. The slowly varying error component mainly corresponds to the long-term thermal memory effect caused by changes in the junction temperature of the power amplifier 10 and fluctuations in the heat sink temperature. Its rate of change is slow, and its spectral energy is mainly concentrated in the low-frequency band, typically below the kilohertz (kHz) level. The rapidly varying error component, on the other hand, mainly corresponds to the short-term electrical memory effect caused by parasitic inductance-capacitance resonance in the internal bias network of the power amplifier 10 and the dynamic response of the notch filter circuit. Its rate of change is extremely fast and closely related to the transient changes in the signal envelope, with its spectral energy typically distributed in the megahertz (MHz) level. Separating these two errors from different sources and with vastly different characteristics is a crucial prerequisite for achieving subsequent targeted and efficient compensation.

[0068] Among them, the slow-varying error component refers to the error component corresponding to the long-term thermal memory effect separated from the total error signal of the power amplifier (mainly caused by changes in device junction temperature and fluctuations in heat sink temperature, with a slow rate of change and spectral energy concentrated below the kHz level). It is used to provide basic data for decoupling compensation control and accurately support the separation and compensation of amplitude distortion and phase distortion.

[0069] The fast-changing error component refers to the error component corresponding to the short-time electrical memory effect separated from the total error signal of the power amplifier (mainly caused by the parasitic inductance and capacitance resonance of the bias network and the dynamic response of the notch filter circuit; the rate of change is extremely fast, the spectral energy distribution is on the order of MHz, and it is closely related to the transient change of the signal envelope). It is used to provide a basis for predictive feedforward control and quickly suppress transient nonlinear distortion.

[0070] Accordingly, both the separated slowly varying error components and the rapidly varying error components are transmitted to the system's brain—the control unit 60. Simultaneously, the input signal of the power amplifier 10 is also sent to the control unit 60 as a reference for feedforward control. The control unit 60 is a powerful processing unit, whose hardware implementation can be a digital signal processor (DSP), a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), or a high-performance microcontroller system composed of these. The control unit 60 receives the separated error components and the input signal, and executes the cooperative linearization algorithm proposed in this application in parallel.

[0071] The control unit 60 executes two main control loops in parallel, which are distinct in time scale and compensation mechanism: one loop is a predictive feedforward control loop based on fast-changing error components (hereinafter referred to as the fast loop). This loop aims to suppress the electrical memory effect at an extremely high speed. The control unit 60 not only passively analyzes the fast-changing error components that have occurred, but more importantly, it actively analyzes the dynamic characteristics of the input signal (such as power transients) in real time to predict the severe distortion that will occur in the power amplifier 10. Based on this predictive analysis, it generates a series of high-precision control commands and sends them to the second compensation unit 40 (also referred to as the feedforward compensation unit or the electrical effect compensation unit). The second compensation unit 40 generates a final cancellation signal according to the commands and injects this signal into one or more bias lines of the power amplifier 10, such as the gate bias line or the drain bias line, through a broadband injection network. By high-speed modulation at the bias point, the dynamic operating point of the transistor can be actively and predictively changed, thereby quickly and effectively suppressing nonlinear distortion caused by the electrical memory effect.

[0072] Among them, predictive feedforward control refers to a control method that actively predicts the distortion that will occur in the power amplifier based on the fast-changing error components and the dynamic characteristics of the input signal and applies a suppression signal in advance (changing passive feedback to active intervention). It is used to quickly respond to transient changes in the signal, suppress short-time electrical memory effects, and improve the dynamic linearity of the power amplifier.

[0073] In parallel, another loop is a decoupling compensation control loop based on the slowly varying error component (hereinafter referred to as the slow loop). This loop aims to stably and accurately compensate for long-term drift caused by thermal effects. The control unit 60 first processes the slowly varying error component, decomposing it into two independent sub-components representing amplitude distortion (gain drift) and phase distortion (phase drift), respectively. Based on the sub-component representing amplitude distortion, the control unit 60 generates an amplitude control command and sends it to the first compensation unit 30 (also referred to as the amplitude compensation unit or thermal effect compensation unit). The first compensation unit 30 injects a compensation current into a power stage (e.g., the drain power node) of the power amplifier 10 according to the command, accurately compensating for gain compression or gain expansion caused by thermal effects by directly adjusting the DC operating point of the power transistor. Simultaneously, based on the sub-component representing phase distortion, the control unit 60 generates a phase control command and sends it to the aforementioned phase pre-correction module 20 located on the input path, accurately compensating for phase drift with temperature by adjusting the pre-correction phase shift applied to the input signal. This amplitude-phase decoupling control method avoids the amplitude and phase crossover effects that may occur with a single compensation method, thus improving the accuracy and stability of compensation.

[0074] Among them, power amplifier 10 refers to the core signal amplification device of this system (usually a high-frequency power amplifier), which is the target of all distortion compensation and is used to amplify the power of the input signal.

[0075] Among them, the phase pre-correction module 20 refers to the high-precision phase shift unit connected in series in the input path of the power amplifier. It is used to receive the phase control command of the control unit 60, apply the pre-correction phase shift to the input signal, and cancel the phase drift of the power amplifier (phase compensation).

[0076] The first compensation unit 30 refers to the current injection unit connected to the power stage of the power amplifier, which is used to receive the amplitude control command of the control unit 60, inject compensation current to adjust the DC operating point of the power tube, and compensate for amplitude distortion (gain drift).

[0077] The second compensation unit 40 is a cancellation signal injection unit connected to the bias line of the power amplifier. It is used to receive the fast loop command from the control unit 60, inject the final cancellation signal, and suppress transient distortion caused by short-time electrical memory effect.

[0078] The error processing unit 50 is used to synchronously acquire the original input signal and the output signal of the power amplifier, calculate the total error signal, and transmit it to the control unit 60.

[0079] The control unit 60 is used to receive signals from the error processing unit, complete logic such as fast / slow error separation, composite transient feature extraction, and model calculation, issue fast / slow loop compensation instructions, and coordinate the work of each unit.

[0080] Among them, decoupling compensation control refers to the control method that orthogonally decomposes the slowly varying error component into in-phase error component and quadrature error component, and then compensates for them separately (breaking the coupling relationship between amplitude distortion and phase distortion).

[0081] The core of the above-mentioned decoupling compensation control lies in first decoupling and separating the amplitude distortion and phase distortion generated by the power amplifier 10 during operation (especially under the effect of long-term thermal memory), breaking the coupling relationship between the two, and then designing compensation strategies for the separated amplitude distortion component and phase distortion component respectively, so as to achieve accurate compensation for the linearity deterioration caused by the long-term thermal memory effect and avoid the problem of neglecting one aspect in traditional coupling compensation.

[0082] Through the closed-loop control system of fast and slow loop working in coordination and amplitude-phase decoupling compensation, this application can comprehensively and efficiently cope with the nonlinear distortion of the power amplifier. It can not only respond quickly to the dynamic distortion caused by signal transients, but also stably correct the performance drift caused by long-term thermal effects.

[0083] Next, refer to Figure 2 The figure is a flowchart of a power amplifier cooperative linearization method provided in an embodiment of this application. The method flow is similar to... Figure 1 The system structures shown correspond one-to-one and represent the logical operation of the system.

[0084] The method includes step S201: error acquisition and separation. As previously described, the system obtains the total error signal by comparing the output sampled signal of the power amplifier 10 with the input reference signal. Then, in the error processing unit 50, the total error signal is separated. As a specific and cost-effective implementation, this separation process can be implemented using an analog filter bank. (Refer to...) Figure 3The total error signal is simultaneously fed into a low-pass filter and a band-pass filter. The low-pass filter, such as a Butterworth low-pass filter with a cutoff frequency of 1 kHz, is used to filter out high-frequency components, thereby extracting the slowly varying error components. The Butterworth filter is chosen because it has the flattest frequency response within its passband, allowing the slowly varying error signal to pass through without distortion. The cutoff frequency is chosen because the time constant of the thermal effect is typically on the order of milliseconds, and its corresponding spectral energy is mainly concentrated in the range from DC to several hundred hertz; a 1 kHz cutoff frequency can effectively retain this part of the signal while filtering out fast-changing components. The band-pass filter, such as a Chebyshev band-pass filter with a passband range of 1 MHz to 10 MHz, is used to extract the fast-changing error components within a specific frequency band. Chebyshev filters were chosen because they exhibit a steeper roll-off characteristic outside the passband, enabling them to more thoroughly filter out slow-change components and out-of-band noise. The passband range was selected because the resonant frequencies formed by parasitic inductance and capacitance in the power amplifier bias network typically fall within this range, which is the primary source of electrical memory effects. This targeted separation method in the frequency domain lays a solid foundation for subsequent targeted compensation.

[0085] After error separation, the method can execute steps S202 and S203 in parallel, reflecting the idea of ​​fast and slow loops working together.

[0086] Step S202 involves executing predictive feedforward control based on rapidly changing error components. The purpose of this step is to transform traditional passive compensation into active intervention, responding to transient changes in the signal at extremely high speeds, thereby nipping distortion in the bud at its earliest stage. Specifically, this step may include the following sub-processes: First, the control unit 60 acquires the input signal of the power amplifier 10 and performs real-time analysis of the envelope power of the input signal to identify power transient events. A power transient event can be defined as the rate of change of the signal envelope power exceeding a preset threshold within a very short time (e.g., 1 microsecond), or a power value jumping above a specific decibel value (e.g., 10 dB) within a short period, a common phenomenon in modern communication signals (such as OFDM and WCDMA). After identifying a power transient event, the control unit 60 extracts composite transient characteristic parameters from the event to quantify the aggressiveness of the transient. Among them, the composite transient characteristic parameters refer to the core parameters (including transient steepness and transient peak power) extracted from the transient events of the input signal power, which are used to quantify the transient intensity. They are used to calculate the dynamic gain factor and provide a quantitative basis for the amplitude modulation of the canceled signal in predictive feedforward control.

[0087] In this embodiment, the composite transient characteristic parameters specifically include transient slew rate and transient peak power. Transient slew rate is defined as the maximum value of the first derivative of the signal envelope power with respect to time during the transient period, reflecting the drastic nature of the power change; transient peak power refers to the maximum power value reached by the signal envelope during the transient period, reflecting the energy level reached by the transient. Accordingly, the control unit 60 calculates a key adjustment factor, namely the dynamic gain factor γ, based on the extracted composite transient characteristic parameters. The dynamic gain factor γ is an adjustment factor calculated based on the composite transient characteristic parameters and used to modulate the amplitude of the fundamental cancellation signal.

[0088] The dynamic gain factor γ can be calculated using the following formula:

[0089] γ=1+k_slew*S_transient+k_peak*(P_peak_transient-P_avg);

[0090] Where γ is the dynamic gain factor; S_transient is the transient kurtosis; P_peak_transient is the transient peak power; k_slew is the preset kurtosis gain coefficient; k_peak is the preset peak power gain coefficient; and P_avg is the average input power over a preset time period (e.g., the past 100 milliseconds). The physical meaning of this formula is that when the signal is stable and there are no significant transients, both S_transient and (P_peak_transient - P_avg) are close to zero. Approximately equal to 1, the control loop performs only routine compensation; however, when a severe power transient occurs, γ will be significantly greater than 1, and its magnitude is positively correlated with the severity of the transient and the peak power. These two gain coefficients, k_slew and k_peak, are adjustable parameters of the system and can be optimized through offline calibration or online adaptive algorithms to achieve the best predictive compensation effect. Simultaneously, the control unit 60 generates a basic cancellation signal based on the fast-changing error component received from the error processing unit 50. Ideally, this basic cancellation signal should be equal in magnitude and opposite in phase to the fast-changing error component. Then, the control unit 60 uses the dynamically calculated dynamic gain factor γ to modulate the amplitude of this basic cancellation signal in real time, thereby obtaining the final cancellation signal. In other words, when a power transient is about to cause severe distortion to the power amplifier 10, the system proactively amplifies the strength of the cancellation signal in advance using a γ value greater than 1 to suppress the expected, more severe distortion. Finally, the control unit 60 controls the second compensation unit 40 to inject the dynamically gain-modulated final cancellation signal into a bias circuit of the power amplifier 10. As a preferred implementation, to improve injection efficiency and compensation effect, this embodiment also provides an adaptive injection point selection method. During system initialization or periodic calibration, the control unit 60 can perform spectral analysis on the noise or injected test signal on the gate bias line and drain bias line of the power amplifier 10 to determine which line the resonant energy caused by the electrical memory effect is mainly distributed on. In actual operation, the control unit 60 will select to inject the final cancellation signal into the bias line where the resonant energy is dominant. This is equivalent to applying compensation energy to the main source of distortion, thereby achieving the most effective distortion suppression.

[0091] Among them, the bias line with dominant resonant energy refers to the bias line in the gate / drain (or base / collector) bias line of the power amplifier, which forms an LC resonant structure by parasitic inductance and parasitic capacitance, and whose resonant energy distribution density is higher than that of other lines (the resonant energy originates from the inherent parasitic parameters of the bias network). It is used to clarify the injection point of the cancellation signal for predictive feedforward control and improve the efficiency of electrical memory effect suppression.

[0092] In existing technologies, the inherent parasitic parameters of the bias network, such as wiring inductance and component parasitic capacitance, form an equivalent LC resonant structure when the amplifier operates at high frequencies. This generates resonant energy, which becomes the dominant characteristic of the bias circuit. This resonant effect causes fluctuations in bias voltage / current, affecting the linearity and operational stability of the power amplifier. Therefore, it is necessary to specifically suppress this effect by injecting a final cancellation signal.

[0093] In parallel with the fast loop, in step S203, decoupling compensation control based on the slowly varying error components is performed. This step aims to accurately and stably compensate for the gain and phase drift caused by long-term thermal effects. Its core lies in avoiding mutual interference between amplitude and phase compensation through decoupling control. The detailed process may include: first, the control unit 60 performs orthogonal decomposition on the received slowly varying error components. (Refer to...) Figure 3 The slowly varying error components are decomposed into an in-phase error component (I component) reflecting gain drift and an orthogonal error component (Q component) reflecting phase drift. The in-phase error component (I component) refers to the independent error component reflecting the power amplifier gain drift (amplitude distortion) after orthogonal decomposition of the slowly varying error components. It is used to provide input for the adaptive temperature compensation model, guiding accurate compensation of amplitude distortion. The orthogonal error component (Q component) refers to the independent error component reflecting the power amplifier phase drift (phase distortion) after orthogonal decomposition of the slowly varying error components. It is used to provide input for the adaptive phase correction model, guiding accurate compensation of phase distortion. Technically, this decomposition process can be achieved by mixing the slowly varying error components with an in-phase (e.g., cosine) local carrier and an orthogonal (e.g., sine) local carrier, followed by strict low-pass filtering, similar to the I / Q demodulation principle in a communication receiver. The resulting in-phase error component is mainly linearly correlated with the amplitude distortion of the power amplifier 10, while the orthogonal error component is mainly linearly correlated with the phase distortion. Next, the control unit 60 performs independent and non-interfering compensation for amplitude and phase based on these two mutually orthogonal and information-decoupled sub-components. For amplitude distortion compensation, refer to... Figure 4 The control unit 60 determines the required compensation current based on the in-phase error component using an adaptive temperature compensation model 70. This adaptive temperature compensation model 70 can be a look-up table (LUT) mapping the in-phase error to the compensation current, or a mathematical function (e.g., a polynomial function) taking the in-phase error (or its integral value) as input. It internally includes at least one temperature compensation coefficient (e.g., ...). Figure 4 The linear coefficients α and nonlinear coefficients β shown define the mapping relationship between the error and the compensation current. The control unit 60 then controls the first compensation unit 30 to inject the calculated compensation current into the power stage (e.g., the drain power node) of the power amplifier 10, precisely compensating for gain drift by fine-tuning the quiescent operating point of the power transistor. For phase distortion compensation, the same principle applies. Figure 4 The control unit 60 determines the required pre-corrected phase shift based on the quadrature error components using an adaptive phase correction model 80. This adaptive phase correction model 80 can also be a lookup table or a mathematical function, containing at least one phase correction coefficient (e.g., ...). Figure 4The linear coefficients δ and nonlinear coefficients ε shown define the relationship between the phase error and the required phase shift. The control unit 60 then controls the phase pre-correction module 20 to apply this pre-correction phase shift to the input signal entering the power amplifier 10, thereby canceling the phase drift it would generate before the signal enters the amplifier. It is understood that, to ensure the accuracy and stability of long-term compensation and to cope with the aging of the power amplifier and environmental changes, both the adaptive temperature compensation model 70 and the adaptive phase correction model 80 can be designed as adaptive models. Specifically, the control unit 60 periodically (e.g., every 100 milliseconds) uses optimization algorithms such as the least mean square (LMS) algorithm or the recursive least squares (RLS) algorithm to iteratively optimize the temperature compensation coefficients (α, β, etc.) in the adaptive temperature compensation model 70 and the phase correction coefficients (δ, ε, etc.) in the adaptive phase correction model 80 based on the accumulated values ​​of the in-phase error component and the quadrature error component. The goal of these algorithms is to adjust the coefficients so that the compensation effect predicted by the model minimizes the long-term accumulated residual error. Through this adaptive mechanism, the system can automatically track and adapt to the slow changes in the characteristics of the power amplifier 10 caused by factors such as device aging and changes in ambient temperature, thus achieving self-calibration.

[0094] Wherein, α is the linear correlation coefficient between temperature change and compensation current, used to characterize the change in basic compensation current corresponding to a unit temperature drift. β is the nonlinear correlation coefficient between temperature change and compensation current, used to correct nonlinear deviations under large temperature drift scenarios. Together, they define the mapping relationship between in-phase error components and compensation current.

[0095] Wherein, δ is the linear correlation coefficient between temperature change and pre-corrected phase shift, used to characterize the basic phase shift compensation amount corresponding to unit temperature drift; ε is the nonlinear correlation coefficient between temperature change and pre-corrected phase shift, used to correct phase nonlinear deviation under large temperature drift scenarios. Together, they define the mapping relationship between orthogonal error components and pre-corrected phase shift.

[0096] Among them, the adaptive temperature compensation model refers to a mathematical model (which can be a lookup table or a polynomial function with built-in temperature compensation coefficients α and β) that maps the in-phase error components to the compensation current. It is used to determine the compensation current required for amplitude compensation based on the in-phase error components, and the coefficients can be iteratively optimized by the least squares method to adapt to device aging and environmental changes.

[0097] An adaptive phase correction model is a mathematical model (which can be a lookup table or a polynomial function with built-in phase correction coefficients δ and ε) that maps orthogonal error components to pre-corrected phase shifts. It is used to determine the pre-corrected phase shifts required for phase compensation based on the orthogonal error components, and the coefficients can be iteratively optimized using the least squares method to ensure long-term phase compensation accuracy.

[0098] In addition, as an optional way to enhance the added value of the system, the method of this embodiment may also include step S204: predictive maintenance information generation.

[0099] Among them, predictive maintenance information refers to the prompt information generated based on the long-term drift trend of key coefficients in the adaptive temperature compensation model and the adaptive phase correction model (such as cumulative drift exceeding the preset threshold), which reflects the aging state of the physical characteristics of the power amplifier. It is used to guide maintenance personnel to inspect or replace components in advance, realizing the transformation from fault repair to predictive maintenance.

[0100] During slow-loop control, the control unit 60 not only uses the coefficients of the adaptive model to perform compensation but also continuously monitors the long-term drift trends of these coefficients. For example, the control unit 60 can record the change trajectory of these coefficients over 1000 hours of continuous operation. If one or more coefficients are found to exhibit a continuous, unidirectional trend with a cumulative drift exceeding a preset threshold (e.g., a change exceeding 20% ​​compared to the initial calibration value), this is usually a clear indication of irreversible aging of the physical characteristics of the power amplifier 10. At this time, the control unit 60 can generate predictive maintenance information about the health status of the power amplifier 10, such as issuing an alarm through system logs, network management interfaces (e.g., SNMP), or panel indicator lights, prompting maintenance personnel that the power amplifier is about to experience performance degradation or potential failure, requiring advance scheduling of maintenance or replacement. This realizes the transformation from traditional reactive maintenance after a fault to proactive, predictive maintenance based on status, improving the reliability and maintainability of the entire communication system.

[0101] Example 2

[0102] This embodiment provides a technical variation of Embodiment 1, the main difference being the specific technical means by which the error processing unit 50 and the control unit 60 implement the error separation step S201, aiming to provide higher separation accuracy, flexibility and stability compared to the simulation scheme.

[0103] In Example 1, error separation was achieved using fixed analog low-pass and band-pass filters. In this example, however, a fully digital processing scheme is employed. Specifically, the error processing unit 50 no longer contains analog filters; instead, it integrates a high-speed, high-precision analog-to-digital converter (ADC). This ADC samples and quantizes the total error signal at a Nyquist sampling rate far exceeding the signal bandwidth (e.g., 200 megasamples / second or higher for a 20MHz bandwidth signal), converting it into a digital signal sequence without distortion.

[0104] The digital signal sequence is directly fed into the control unit 60. In this embodiment, the control unit 60 is preferably a field-programmable gate array (FPGA) or a dedicated digital signal processor with powerful digital signal processing (DSP) capabilities. Within the digital domain, the control unit 60 implements highly flexible digital filter algorithms through software programming or hardware logic to perform error separation. For example, the control unit 60 can implement a digital low-pass filter (such as a finite-length unit impulse response (FIR) filter or an infinite-length unit impulse response (IIR) filter) to extract slowly varying error components, and simultaneously implement a digital band-pass filter to extract rapidly varying error components.

[0105] Understandably, this fully digital processing method offers significant advantages over analog filters. First, all parameters of a digital filter (such as cutoff frequency, passband width, order, roll-off factor, etc.) can be easily configured and dynamically adjusted via software. This allows the system to more accurately match the spectral characteristics of different power amplifier models or different communication signal standards, and even adaptively adjust online to achieve optimal error separation. Second, digital filters can achieve a more ideal frequency response than analog filters, such as a steeper transition band, a flatter passband, and higher stopband attenuation, resulting in less crosstalk between fast and slow-varying error components and better separation. Furthermore, the digital implementation avoids the temperature drift and aging problems of analog components, improving the long-term stability and consistency of the system. It should be noted that the superior error separation provides a cleaner input signal for subsequent predictive feedforward control and decoupling compensation control, directly translating into improved overall linearization performance.

[0106] Apart from the specific implementation of error separation, the principles and processes of subsequent steps S202 (predictive feedforward control), S203 (decoupling compensation control), and S204 (predictive maintenance) in this embodiment are basically the same as those in Embodiment 1, except that their input signals (fast / slow variable error components) come from the results of digital domain processing.

[0107] Example 3

[0108] This embodiment illustrates that the technical solution proposed in this application has good versatility and device independence, and can be applied to power amplifiers manufactured using different semiconductor processes. Embodiment 1 mainly uses a field-effect transistor (FET) type power amplifier as an example, with its bias circuit consisting of a gate and a drain. This embodiment applies this technical solution to a power amplifier based on a bipolar junction transistor (BJT) or heterojunction bipolar transistor (HBT), which are also common in some applications.

[0109] The overall structure of the system (such as) Figure 1(as shown) and core methodological processes (such as) Figure 2 (As shown) This embodiment is consistent with Example 1. The core technical concept of this application, namely error separation based on physical mechanisms, fast-loop predictive feedforward, and slow-loop decoupling compensation, is a universal control theory that does not depend on a specific transistor type. The main adjustment in this embodiment lies in the selection of the compensation signal injection point to match the physical structure and working principle of the BJT or HBT device.

[0110] Specifically, adaptive adjustments may include injecting a final cancellation signal into the bias line to suppress the electrical memory effect during the predictive feedforward control in step S202. For BJT or HBT devices, the corresponding control and power terminals are the base and collector, respectively. Accordingly, during the spectral analysis for injection point selection, the analysis should focus on the signals on the base bias line and collector bias line to determine the distribution of resonant energy. Finally, the second compensation unit 40 injects the final cancellation signal into the bias line (i.e., the base bias line or the collector bias line) where the resonant energy is dominant, as determined by the analysis.

[0111] When performing the amplitude compensation section of the decoupling compensation control in step S203, a compensation current needs to be injected into the power stage to compensate for thermal effects. For BJT or HBT devices, the primary power control node is the collector. Therefore, the injection point of the slowly varying compensation current generated by the first compensation unit 30 for compensating gain drift should be adjusted to the collector node of the final stage power transistor of the power amplifier 10.

[0112] Through the simple adaptive adjustments to the physical structure of the devices described above, the cooperative linearization method and system disclosed in this application can be seamlessly applied to various mainstream power amplifier technologies, achieving good linearization results for both field-effect transistors and bipolar transistors.

[0113] Example 4

[0114] This embodiment provides a high-performance hardware implementation scheme for the control unit 60, which is applicable to demanding application scenarios requiring the processing of ultra-wideband, high peak-to-average power ratio (PAPR) signals, such as 5G NR (New Radio) communication systems or broadband radar systems. As an optional implementation, the control unit 60 can employ a heterogeneous hybrid control architecture. Figure 1 The control unit 60 provides a specific, high-performance internal structure to achieve an optimal balance between speed, complexity, and cost.

[0115] In this embodiment, the control unit 60 employs a hybrid control architecture, internally divided into two cooperating sub-units, each implemented by a different type of processor. The first sub-unit is a high-speed control loop unit, or fast loop processor. Leveraging the inherent massively parallel processing capabilities and extremely low logic latency of field-programmable gate arrays (FPGAs), this unit is implemented using a single FPGA and is specifically designed to execute a real-time-critical predictive feedforward control loop (i.e., fast loop, step S202). Specifically, this FPGA unit is directly connected to the error processing unit 50 (or its internally integrated ADC), receiving the high-speed sampled input signal and the total error signal. Utilizing its parallel hardware logic, it performs real-time identification of power transient events in the input signal, extraction of composite transient characteristic parameters, calculation of the dynamic gain factor γ, and modulation of the basic cancellation signal within nanoseconds or submicroseconds. After calculation, the FPGA unit directly generates digital or analog control signals to drive the second compensation unit 40 to inject the cancellation signal. All these operations are completed within one or several clock cycles, ensuring a rapid response to signal transient distortion.

[0116] The second sub-unit is the low-speed control and management loop unit, also known as the slow loop and management processor. This unit is implemented by a general-purpose microcontroller unit (MCU) or a digital signal processor (DSP) and is responsible for handling relatively complex logic but low real-time requirements (typically on the order of milliseconds) decoupling compensation control loops (i.e., slow loops, step S203) and system management tasks (such as step S204). Specifically, this MCU / DSP unit receives the slowly varying error components separated by digital filtering from the FPGA unit. Then, it performs orthogonal decomposition of the slowly varying error components in software, runs the adaptive temperature compensation model 70 and the adaptive phase correction model 80, calculates the compensation amount, and controls the first compensation unit 30 and the phase pre-correction module 20, respectively. In addition, computationally intensive or management tasks such as periodic model coefficient iterative optimization (e.g., running computationally intensive LMS or RLS algorithms) and monitoring, analysis, and generation of predictive maintenance information are also undertaken by this MCU / DSP unit.

[0117] Reference Figure 5 This diagram illustrates the interaction timing of key signaling under the hybrid control architecture of this embodiment, clearly demonstrating the advantages of this hardware and software task partitioning. In a typical control cycle, after startup at time t0, the FPGA completes the acquisition of a new frame of signals at time t1, and completes digital domain error separation at time t2. Simultaneously, at time t2, the FPGA communicates via the internal bus, SPI, or I / O. 2The C-type interface sends the slowly varying error component to the MCU / DSP. Afterward, the FPGA and MCU / DSP enter a parallel processing phase. The FPGA focuses on fast-loop control, while the MCU / DSP focuses on slow-loop control and system management. Specifically, the FPGA utilizes its hardware parallelism to complete all fast-loop calculations in an extremely short time (e.g., from t2 to t3, the time difference is less than 1 microsecond). After completing the calculation at time t3, it immediately sends a compensation command to the second compensation unit 40 at time t4, thus ensuring an extremely rapid response to transient signal distortions. This response time is much smaller than the signal envelope's variation period. Simultaneously, the MCU / DSP receives the slowly varying error component after t2 and begins executing its software calculation tasks, such as orthogonal decomposition, model computation, and coefficient optimization. It should be noted that since the slow loop is designed to handle slowly changing thermal effects, its processing delay (e.g., up to several milliseconds) does not adversely affect the performance of the fast loop; the two are completely decoupled in time.

[0118] By assigning fast and slow loop tasks to the most suitable processors (i.e., using an FPGA for speed and parallelism, and an MCU / DSP for complex logic and flexibility) in a hybrid control architecture, this embodiment can balance high-speed dynamic response and complex algorithm processing capabilities while keeping costs under control, providing a solid and feasible hardware foundation for the application of the technical solution in this application in future high-performance systems such as communications.

[0119] The above descriptions are merely several preferred embodiments of this application and are not intended to limit the scope of protection of this application. For those skilled in the art, various modifications, alterations, substitutions, and variations can be made to the above embodiments without departing from the spirit and principles of the technical solutions of this application. For example, technical features in different embodiments can be combined arbitrarily, as long as no contradiction arises. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection claimed by the claims of this application.

Claims

1. A method for cooperative linearization of a power amplifier, characterized in that, The method includes: Obtain the error signal that reflects the output distortion of the power amplifier, and separate the error signal into a slow-changing error component and a fast-changing error component; Predictive feedforward control based on the fast-changing error components is performed to suppress the short-time electrical memory effect of the power amplifier; Perform decoupling compensation control based on the slowly varying error component to compensate for the long-term thermal memory effect of the power amplifier; The execution of predictive feedforward control based on the rapidly changing error components includes: The system acquires an input signal; analyzes the input signal to identify power transient events and extracts composite transient characteristic parameters from the power transient events; calculates a dynamic gain factor based on the composite transient characteristic parameters; generates a basic cancellation signal based on the fast-changing error component; modulates the amplitude of the basic cancellation signal using the dynamic gain factor to obtain a final cancellation signal; and injects the final cancellation signal into a bias line of the power amplifier. The composite transient characteristic parameters include transient kurtosis and transient peak power; The calculation of the dynamic gain factor based on the composite transient characteristic parameters includes: The dynamic gain factor is calculated using the following formula: γ=1+k_slew*S_transient+k_peak*(P_peak_transient-P_avg); Wherein, γ is the dynamic gain factor; S_transient is the transient steepness; P_peak_transient is the transient peak power; k_slew is the preset steepness gain coefficient; k_peak is the preset peak power gain coefficient; and P_avg is the average input power within a preset time period. The execution of decoupling compensation control based on the slowly varying error component includes: The slowly varying error component is orthogonally decomposed into an in-phase error component reflecting gain drift and an orthogonal error component reflecting phase drift; The amplitude distortion of the power amplifier is compensated based on the in-phase error component. The phase distortion of the power amplifier is compensated based on the orthogonal error components.

2. The power amplifier cooperative linearization method as described in claim 1, characterized in that, The step of separating the error signal into a slowly varying error component and a rapidly varying error component includes: The slowly varying error component is extracted from the error signal using a low-pass filter; The rapidly changing error component is extracted from the error signal using a bandpass filter.

3. The power amplifier cooperative linearization method as described in claim 1, characterized in that, The bias lines include gate bias lines and drain bias lines; The process includes, before injecting the final cancellation signal into a bias line of the power amplifier: Spectral analysis is performed on the signals on the gate bias line and drain bias line of the power amplifier to determine the distribution of resonant energy; The step of injecting the final cancellation signal into a bias line of the power amplifier includes: injecting the final cancellation signal into a bias line where the resonant energy is dominant, based on the distribution of the resonant energy.

4. The power amplifier cooperative linearization method as described in claim 1, characterized in that, The compensation for amplitude distortion of the power amplifier based on the in-phase error component includes: An adaptive temperature compensation model is used to determine the compensation current and perform amplitude compensation. Based on the cumulative value of the in-phase error component, a least squares optimization algorithm is used to iteratively optimize at least one temperature compensation coefficient in the adaptive temperature compensation model. The compensation for phase distortion of the power amplifier based on the orthogonal error components includes: An adaptive phase correction model is used to determine the pre-corrected phase shift and perform phase compensation; based on the cumulative value of the orthogonal error components, a least squares optimization algorithm is used to iteratively optimize at least one phase correction coefficient in the adaptive phase correction model.

5. The power amplifier cooperative linearization method as described in claim 4, characterized in that, The method further includes: Monitor the long-term drift trends of the temperature compensation coefficient in the adaptive temperature compensation model and the phase correction coefficient in the adaptive phase correction model; Based on the long-term drift trend, predictive maintenance information about the health status of the power amplifier is generated.

6. A power amplifier cooperative linearization system, characterized in that, include: An error processing unit is used to acquire an error signal and separate the error signal into a slowly varying error component and a rapidly varying error component. A feedforward compensation unit is used to generate a cancellation signal and inject the cancellation signal into the bias circuit of the power amplifier; The first compensation unit is connected to a power stage of the power amplifier and is used to compensate for the amplitude distortion of the power amplifier. A phase pre-correction module is connected in series in one of the input signal paths of the power amplifier to compensate for the phase distortion of the power amplifier. The control unit, connected to the error processing unit, is used to acquire the input signal of the power amplifier, receive the slow-varying error component and the fast-varying error component, and is used to: control the feedforward compensation unit to perform predictive feedforward control based on the fast-varying error component; and control the first compensation unit and the phase pre-correction module to perform decoupling compensation control based on the slow-varying error component. The execution of predictive feedforward control based on the rapidly changing error components includes: The system acquires an input signal; analyzes the input signal to identify power transient events and extracts composite transient characteristic parameters from the power transient events; calculates a dynamic gain factor based on the composite transient characteristic parameters; generates a basic cancellation signal based on the fast-changing error component; modulates the amplitude of the basic cancellation signal using the dynamic gain factor to obtain a final cancellation signal; and injects the final cancellation signal into a bias line of the power amplifier. The composite transient characteristic parameters include transient kurtosis and transient peak power; The calculation of the dynamic gain factor based on the composite transient characteristic parameters includes: The dynamic gain factor is calculated using the following formula: γ=1+k_slew*S_transient+k_peak*(P_peak_transient-P_avg); Wherein, γ is the dynamic gain factor; S_transient is the transient steepness; P_peak_transient is the transient peak power; k_slew is the preset steepness gain coefficient; k_peak is the preset peak power gain coefficient; and P_avg is the average input power within a preset time period. The implementation of decoupling compensation control based on the slowly varying error component includes: The slowly varying error component is orthogonally decomposed into an in-phase error component reflecting gain drift and an orthogonal error component reflecting phase drift; The amplitude distortion of the power amplifier is compensated based on the in-phase error component. The phase distortion of the power amplifier is compensated based on the orthogonal error components.

7. The power amplifier cooperative linearization system as described in claim 6, characterized in that, The control unit adopts a hybrid control architecture, including: The high-speed control loop unit, implemented by a field-programmable gate array (FPGA), is used to receive the input signal and the fast-changing error component, extract the composite transient characteristic parameters and calculate the dynamic gain factor in real time, and control the feedforward compensation unit. The low-speed control and management loop unit, implemented by a microcontroller unit (MCU), is used to process the slowly varying error components and perform their orthogonal decomposition, thereby controlling the first compensation unit and the phase pre-correction module, respectively.

Images