Design method of broadband microwave solid-state power amplifier based on reconfigurable matching network

Abstract

The application discloses a broadband microwave solid-state power amplifier design method based on a reconfigurable matching network, and belongs to the technical field of microwave power amplifier design. The method acquires transistor model parameters and a target working frequency band, constructs an initial topology of a reconfigurable matching network containing tunable element nodes and switch switching paths, performs discretization sampling on the target frequency band to obtain a plurality of key frequency points, dynamically adjusts element values and switch on-off states for each key frequency point through an iterative search algorithm to generate a corresponding power matching parameter set, compares the power matching parameter sets of adjacent key frequency points in terms of difference degree, identifies parameter jump intervals with a difference degree exceeding a tolerance threshold, positions mismatch element nodes according to the parameter jump intervals, performs a switch path reconstruction operation, and generates a broadband matching parameter sequence. The application can effectively solve the performance deterioration problem of a broadband power amplifier matching network caused by parameter mutation when working across frequency bands.

Description

Technical Field

[0001] This invention relates to the field of microwave power amplifier design technology, specifically to a design method for broadband microwave solid-state power amplifiers based on reconfigurable matching networks. Background Technology

[0002] The demand for broadband microwave solid-state power amplifiers is increasing in systems such as radar, communications, and electronic warfare. To achieve high output power and efficiency across a wide bandwidth, the design of the matching network is crucial. Existing broadband power amplifier matching networks often employ fixed-topology lumped-parameter or distributed-parameter networks, achieving approximate in-band matching through multi-section impedance transformation or filter synthesis. However, these schemes are constrained by the gain-bandwidth product, making it difficult to simultaneously achieve high gain, high efficiency, and a flat power response across the entire target frequency band. Another approach introduces tunable components such as varactor diodes or ferroelectric capacitors, achieving limited-range dynamic matching through bias voltage adjustment. However, this adjustment dimension is singular; when the operating frequency spans multiple octaves, the intrinsic impedance of the transistor changes drastically with frequency. Fixed topologies or simple tuning structures are prone to severe impedance mismatch in specific frequency bands, leading to power gain dips, oscillations, or a sharp drop in efficiency.

[0003] Existing dynamic matching methods generally rely on pre-calculated lookup tables or continuous tuning curves, lacking the ability to locate specific mismatch nodes in the matching network and perform targeted reconfiguration. When the required matching parameters differ significantly at different frequencies within a frequency band, a single tuning element often cannot simultaneously meet the impedance transformation requirements of adjacent frequency bands, resulting in performance degradation within the parameter transition range. Furthermore, existing optimization design processes typically involve continuous scanning of the frequency band or calculation using analytical formulas, failing to combine discretized key frequencies, multi-dimensional switching element states, and tolerance-aware mismatch localization mechanisms. This makes it difficult to systematically generate a wide-bandwidth matching parameter sequence to drive the reconfigurable network to operate smoothly.

[0004] Therefore, it is necessary to solve the problem of how to automatically identify the matching failure section caused by parameter jumps in the design of broadband reconfigurable matching networks, locate the specific component node that causes the mismatch in the failure section and reconfigure the switching path; at the same time, it is also necessary to solve the problem of how to jointly search for multidimensional continuous component values ​​and discrete switch states at discrete key frequency points in order to efficiently obtain the optimal parameter combination that meets the power matching requirements of each frequency point. Summary of the Invention

[0005] This paper presents a design method for broadband microwave solid-state power amplifiers based on reconfigurable matching networks. By performing joint iterative optimization of tunable component values ​​and switching states at discretized key frequency points, and automatically identifying parameter jump segments and mismatched component nodes based on the comparison of matching parameter differences between adjacent frequency points, a switching path reconstruction is performed to generate a broadband matching parameter sequence covering the entire target frequency band. This overcomes the impedance mismatch problem of fixed topology matching networks and simple tuning mechanisms in broadband applications, and improves the gain flatness and design efficiency of broadband power amplifiers.

[0006] The objective of this invention can be achieved through the following technical solutions: This invention provides a design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network, comprising the following steps: Obtain the transistor model parameters of the power amplifier and the preset target operating frequency band range; An initial topology of the reconfigurable matching network is constructed based on the transistor model parameters. The initial topology includes multiple tunable element nodes and multiple switch-switching paths. Discretized frequency sampling processing is performed on the target operating frequency band to obtain multiple key frequency points; For each key frequency point, the component values ​​of the tunable element nodes and the on / off state of the switch switching path are dynamically adjusted through an iterative search algorithm to generate a set of power matching parameters corresponding to each key frequency point. The power matching parameter sets corresponding to adjacent key frequency points are compared to identify parameter jump intervals where the difference exceeds the preset tolerance threshold. Based on the parameter jump interval, the mismatched element node in the reconfigurable matching network is located, and a switch path reconstruction operation is performed on the mismatched element node to generate a broadband matching parameter sequence.

[0007] Through the above steps, the present invention performs reconfigurable design of the matching network in the broadband range, can optimize matching parameters at multiple frequency points, and automatically identify and reconstruct matching mismatch nodes caused by bandwidth expansion, thereby generating an efficient matching parameter sequence covering the entire broadband, significantly improving the gain flatness and power efficiency of the power amplifier in the entire operating frequency band.

[0008] As a preferred embodiment of the present invention, the process of constructing the initial topology of the reconfigurable matching network based on the transistor model parameters specifically involves: analyzing the parasitic capacitance and inductance values ​​in the transistor model parameters to obtain the intrinsic impedance characteristics of the transistor; determining the impedance transformation target values ​​of the input and output ports of the matching network based on the intrinsic impedance characteristics of the transistor; selecting multiple inductor nodes and multiple capacitor nodes, and connecting the inductor nodes and capacitor nodes in series or parallel to form multiple matching branches; embedding an RF switch element in each matching branch as the switching path, connecting the first end of the RF switch element to the input end of the inductor node or the capacitor node, and connecting the second end of the RF switch element to a spare matching node. This constructed initial topology possesses both continuous adjustability and discrete switching capability, providing sufficient adjustment freedom for subsequent broadband matching.

[0009] Furthermore, a preferred method for performing discretized frequency sampling processing on the target operating frequency band is as follows: Obtain the lower limit frequency value and the upper limit frequency value of the target operating frequency band; divide the frequency band interval between the lower limit frequency value and the upper limit frequency value into multiple candidate frequency points according to a logarithmic frequency scale; for each candidate frequency point, calculate the frequency ratio between the candidate frequency point and its adjacent candidate frequency points; determine candidate frequency points whose frequency ratio is less than a preset uniformity threshold as redundant frequency points and remove them, retaining the remaining candidate frequency points as the key frequency points. This sampling method can effectively characterize broadband characteristics with the fewest key frequency points, reducing the computational load of subsequent optimization.

[0010] Furthermore, the iterative search algorithm preferably employs a particle swarm optimization algorithm or a genetic algorithm as a global optimization engine to enhance the ability to find the optimal matching parameters in a multidimensional parameter space.

[0011] As a specific embodiment of the present invention, the step of generating a power matching parameter set for each key frequency point includes: for the current key frequency point, using the current component value of the tunable element node as the search starting point, constructing a multi-dimensional parameter search space; randomly generating multiple candidate parameter vectors within the multi-dimensional parameter search space, each candidate parameter vector containing a combination of component values ​​of all tunable element nodes and the on / off state codes of all switching paths; inputting each candidate parameter vector into an electromagnetic simulator for power gain calculation to obtain the predicted power gain value corresponding to each candidate parameter vector; selecting the candidate parameter vector with the highest predicted power gain value as the elite parameter vector, and re-delineating the search neighborhood with the elite parameter vector as the center, generating a new generation of candidate parameter vectors within the search neighborhood; repeating the steps of power gain calculation, elite selection, and search neighborhood reduction until the change in the predicted power gain values ​​of multiple generations of elite parameter vectors is less than a preset convergence threshold, and using the current elite parameter vector as the power matching parameter set corresponding to the current key frequency point. This search method can quickly converge to the globally optimal matching state, taking into account the comprehensive optimization of continuous element tuning and discrete switching.

[0012] In the above scheme, the electromagnetic simulator preferably uses a harmonic balance simulation engine to calculate the predicted power gain value, so as to accurately reflect the nonlinear effect of the microwave power amplifier and ensure the engineering availability of the matching parameters.

[0013] As another preferred technical solution of the present invention, the specific process of the difference comparison processing is as follows: starting from the first key frequency point, extract the first power matching parameter set of the current key frequency point and the second power matching parameter set of the next key frequency point; calculate the absolute difference between the component values ​​of each tunable element node at the same position in the first power matching parameter set and the second power matching parameter set to obtain the component value difference sequence; count the number of element nodes in the component value difference sequence whose absolute difference exceeds a preset component change threshold, and use the ratio of the number of element nodes to the total number of element nodes as the difference degree; compare the difference degree with the preset tolerance threshold, and if the difference degree is greater than the preset tolerance threshold, mark the frequency band interval between the current key frequency point and the next key frequency point as the parameter jump interval. Through this quantitative comparison, the interval where the matching parameters between adjacent frequency points change abruptly can be accurately captured, providing a clear target for subsequent reconstruction.

[0014] The preferred steps for locating mismatched component nodes based on the parameter transition interval include: extracting a first power matching parameter set and a second power matching parameter set corresponding to the parameter transition interval; comparing the component values ​​in the first power matching parameter set with the component values ​​in the second power matching parameter set node by node, and selecting the top few tunable component nodes with the largest component value changes as candidate mismatched nodes; for each candidate mismatched node, tracking the on / off state change records of the switch switching path in the matching branch where the candidate mismatched node is located; if the on / off state change records show that the on / off state of the RF switch element in the matching branch has not changed before and after the parameter transition interval, then the candidate mismatched node is confirmed as the mismatched component node. This location method eliminates parameter change interference caused by active switch switching and accurately identifies mismatched components whose parameters cannot follow due to frequency changes.

[0015] Further, the step of performing a switch path reconstruction operation on the mismatched element node to generate a broadband matching parameter sequence includes: obtaining the current on / off state of all RF switch elements in the current matching branch where the mismatched element node is located; traversing the backup switch paths containing backup matching element nodes in the current matching branch, and querying the alternative element value combination corresponding to each backup switch path; substituting the alternative element value combination of each backup switch path into the electromagnetic simulator, calculating the power gain at multiple test frequency points within the parameter transition interval, and obtaining the bandwidth gain flatness index of each backup switch path; selecting the backup switch path with the optimal bandwidth gain flatness index as the target reconstruction path, setting the RF switch elements in the target reconstruction path to the on state, and setting the RF switch elements in the original matching branch to the off state; updating the element values ​​of the mismatched element node and the element values ​​of adjacent tunable element nodes, generating a new power matching parameter set covering the parameter transition interval, and concatenating the power matching parameter set corresponding to each key frequency point with the new power matching parameter set in frequency order to obtain the broadband matching parameter sequence. This reconfiguration method effectively suppresses parameter jumps and achieves a smooth transition of gain characteristics within a wide bandwidth by switching the entire switching path and coordinating the adjustment of component values.

[0016] The frequency band gain flatness index is preferably obtained by calculating the difference between the maximum and minimum power gain at each test frequency point within the parameter jump interval. The smaller the difference, the better the flatness.

[0017] As a further improvement of the present invention, after locating the mismatched element node, the following steps are included: extracting the first operating frequency response curve of the mismatched element node before the parameter jump interval and the second operating frequency response curve after the parameter jump interval; calculating the amplitude difference curve between the first operating frequency response curve and the second operating frequency response curve; identifying the frequency point from the amplitude difference curve where the amplitude difference exceeds a preset resonance offset threshold as the resonance drift frequency point; determining the parasitic parameter compensation direction of the mismatched element node based on the resonance drift frequency point, wherein the parasitic parameter compensation direction is used to indicate whether to increase or decrease the parallel compensation capacitor value of the mismatched element node. In this way, it is possible to accurately determine whether the mismatch is caused by the resonance point shift due to parasitic effects, and to compensate accordingly.

[0018] Based on this, the switching path reconstruction operation for the mismatched element node further includes: selecting matching compensation capacitor elements from a pre-built compensation element library according to the parasitic parameter compensation direction; connecting the compensation capacitor elements in parallel to both ends of the mismatched element node through additional RF switching elements to form a compensation branch; adjusting the conduction time sequence of the additional RF switching elements in the compensation branch so that the compensation capacitor elements are connected within the frequency range corresponding to the parameter transition interval and disconnected within the frequency range outside the parameter transition interval; adding the access state code of the compensation branch to the power matching parameter set and updating the equivalent element value of the mismatched element node. Through the controllable access of the parasitic compensation branch, parasitic effects in different frequency bands can be dynamically compensated, further improving matching accuracy and bandwidth performance.

[0019] As a preferred embodiment of the present invention, after generating the broadband matching parameter sequence, the method further includes the following steps: arranging each power matching parameter set in the broadband matching parameter sequence according to the frequency order of key frequency points to form a parameter configuration time axis; identifying the switching time points between two adjacent power matching parameter sets on the parameter configuration time axis; inserting a switching protection time window at each switching time point, the length of which is equal to the sum of the maximum operating time of the RF switching element in the reconfigurable matching network and a preset stabilization margin time; and compiling the parameter configuration time axis after inserting the switching protection time window into an executable switch drive instruction sequence, which is used to sequentially control the on / off state switching of each RF switching element in the reconfigurable matching network. By inserting a switching protection time window, it can be ensured that the parameters take effect only after the switching action is completed, avoiding the impact of switching transients on the power amplifier performance and ensuring the stability and reliability during broadband operation.

[0020] The beneficial effects of this invention are: By comparing the differences in power matching parameter sets corresponding to adjacent key frequency points, this method identifies parameter jump intervals where the differences exceed a preset tolerance threshold. Based on these intervals, it locates mismatched element nodes in the reconfigurable matching network and performs switching path reconstruction operations on these nodes, generating a broadband matching parameter sequence. This approach automatically captures parameter abrupt change segments based on discrete frequency point optimization results, attributing drastic changes in element values ​​in the matching network to specific mismatched nodes. By switching backup switching paths to reconstruct local matching branches, it overcomes the limitation of fixed topology networks that cannot adapt to impedance abrupt changes in a timely manner through continuous tuning. This enables precise compensation for concentrated impedance mismatch areas within the frequency range, allowing for segmented adaptation of the impedance trajectory across the entire broadband range, effectively reducing gain fluctuations and expanding usable bandwidth.

[0021] By discretizing the target operating frequency band and sampling multiple key frequencies, an iterative search algorithm is used for each key frequency to dynamically adjust the component values ​​of the tunable element nodes and the on / off states of the switching paths in a multidimensional parameter search space, generating a set of power matching parameters corresponding to each frequency. This process transforms the broadband matching problem in the continuous frequency dimension into a multidimensional mixed-integer programming problem in a finite number of discrete frequency points. The algorithm searches for the globally optimal matching parameters in the joint space of continuous component values ​​and discrete switch states, avoiding the omissions in the solution space caused by point-by-point scanning or analytical approximation. Thus, even when the transistor impedance changes significantly with frequency, matching parameters that balance optimal gain and the simplest switch combination can be obtained at each key frequency point, providing a high-confidence frequency matching benchmark for subsequent parameter jump analysis and reconstruction, thereby supporting the generation of broadband matching parameter sequences. Attached Figure Description

[0022] The invention will now be further described with reference to the accompanying drawings.

[0023] Figure 1 This is a flowchart of a broadband microwave solid-state power amplifier design method based on reconfigurable matching networks; Figure 2 This is a flowchart of the initial topology construction and key frequency point determination for a reconfigurable matching network; Figure 3 This is a flowchart of the iterative search process for power matching parameters based on particle swarm optimization; Figure 4 This is a flowchart of a method for reconfigurable matching network mismatch compensation and switch drive instruction generation. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] See Figure 1 This invention provides a design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network, comprising the following steps: obtaining the transistor model parameters of the power amplifier and a preset target operating frequency band; constructing an initial topology of the reconfigurable matching network based on the transistor model parameters, the initial topology containing multiple tunable element nodes and multiple switching paths; performing discretization sampling processing on the target operating frequency band to obtain multiple key frequencies; for each key frequency, dynamically adjusting the element values ​​of the tunable element nodes and the on / off state of the switching paths through an iterative search algorithm to generate a power matching parameter set corresponding to each key frequency; performing difference comparison processing on the power matching parameter sets corresponding to adjacent key frequencies to identify parameter jump intervals where the difference exceeds a preset tolerance threshold; locating mismatched element nodes in the reconfigurable matching network based on the parameter jump intervals, and performing switching path reconstruction operations on the mismatched element nodes to generate a broadband matching parameter sequence.

[0026] In specific implementation, please refer to Figure 2 The process of constructing the initial topology of a reconfigurable matched network based on transistor model parameters is as follows: Parasitic capacitance and inductance values ​​in the transistor model parameters are analyzed to obtain the intrinsic impedance characteristics of the transistor. Transistor model parameters can be extracted from simulation model files provided by the transistor manufacturer or from S-parameters measured using a vector network analyzer. The extraction process involves biasing the transistor at a preset DC operating point, measuring the small-signal S-parameters, removing the influence of the test fixture using de-embedding techniques, converting the S-parameters into Y or Z parameters, and separating parasitic element values ​​such as gate-source parasitic capacitance, drain-source parasitic capacitance, gate parasitic inductance, and drain parasitic inductance from the transistor's equivalent circuit topology. Based on the separated parasitic capacitance and inductance values, the intrinsic impedance or admittance parameters of the transistor are obtained by subtracting the influence of parasitic elements from the measured or simulated overall port parameters, thus forming the intrinsic impedance characteristics of the transistor.

[0027] The target impedance transformation values ​​for the input and output ports of the matching network are determined based on the intrinsic impedance characteristics of the transistor. In specific microwave solid-state power amplifier designs, the target impedance transformation value for the input port is determined according to the maximum gain matching or optimal noise matching conditions; the target impedance transformation value for the output port is determined according to the optimal load impedance corresponding to the maximum output power or highest efficiency obtained from load pull measurements. All target impedance transformation values ​​are expressed in complex impedance form, including both real resistance and imaginary reactance components.

[0028] Multiple inductor and capacitor nodes are selected and connected in series or parallel to form multiple matching branches. The initial component values ​​and topology of the inductor and capacitor nodes are determined by transmission line transformers, lumped parameter LC ladder networks, or hybrid structures based on the impedance transformation ratio and operating frequency band. In low microwave frequencies, lumped inductors and lumped capacitors are preferred to form π-type or T-type matching networks; in higher microwave frequencies, microstrip lines or coplanar waveguides are used to form equivalent inductor and capacitor nodes. The resulting multiple matching branches include at least one combination of series inductor branches, parallel capacitor branches, series capacitor branches, and parallel inductor branches.

[0029] An RF switch element is embedded in each matching branch as a switching path. The first end of the RF switch element is connected to the input of an inductor node or capacitor node, and the second end is connected to a backup matching element node. The RF switch element uses a PIN diode switch, a gallium arsenide pseudocrystalline high electron mobility transistor switch, or a microelectromechanical system (MEMS) switch. Several backup inductors or capacitors with different nominal values ​​are pre-set on the backup matching element node. When the RF switch element is in the ON state, the original matching branch is disconnected, and the signal flows through the RF switch element to the backup matching element node, thereby changing the effective element values ​​and topology of the matching network. Initially, all RF switches are set to OFF, and the signal only flows through the matching branch formed by the initially selected inductor and capacitor nodes.

[0030] The process of performing discretized frequency sampling on the target operating frequency band to obtain multiple key frequency points involves acquiring the lower and upper frequency limits of the target operating frequency band. These lower and upper frequency limits are given by the power amplifier's design specifications; for example, the target operating frequency band ranges from... arrive .

[0031] The frequency band between the lower and upper frequency limits is divided into equal intervals according to a logarithmic frequency scale to obtain multiple candidate frequency points. The candidate frequency points are determined by the following formula: in, Indicates the first The frequency values ​​of the candidate frequency points This represents the lower limit frequency value of the target operating frequency band. This represents the upper limit frequency value of the target operating frequency band. This indicates the total number of candidate frequency points pre-defined within the frequency band. The value is an integer not less than 20. From 1 to The candidate frequencies are sequentially incremented integers. The spacing between each adjacent candidate frequency point on the logarithmic frequency axis is equal, resulting in a denser distribution of candidate frequency points in the low-frequency band and a sparser distribution in the high-frequency band, which is consistent with the nonlinear characteristics of the frequency response of microwave devices.

[0032] For each candidate frequency point, calculate the frequency ratio between the candidate frequency point and its adjacent candidate frequency points. The frequency ratio is calculated as follows: for the ... The candidate frequency point and the first Candidate frequency points, frequency ratio Represented as Candidate frequency points with a frequency ratio less than a preset uniformity threshold are identified as redundant and eliminated, while the remaining candidate frequency points are retained as key frequency points. The preset uniformity threshold is set based on the gain flatness requirements of the power amplifier within its operating frequency band, and its value ranges from 1.05 to 1.20. If the frequency ratio between adjacent candidate frequency points is less than a preset uniformity threshold, the candidate frequency points are considered redundant and eliminated. A value close to 1 indicates that the two candidate frequency points are too close, and the matching network parameters change only slightly within this frequency range. Retaining one of them is sufficient to characterize the matching properties of this range. Therefore, the candidate frequency point with the larger index is selected. Redundant frequency points are identified and eliminated. After comparing the frequency ratios of all candidate frequency points and eliminating redundant frequency points, the remaining candidate frequency points constitute the key frequency point sequence. The frequency ratios between adjacent frequency points in the key frequency point sequence are not less than a preset uniformity threshold, and the key frequency points completely cover the lower and upper frequency values ​​of the target operating frequency band.

[0033] In its implementation, the iterative search algorithm employs Particle Swarm Optimization (PSO) as its global optimization engine. The core architecture of PSO includes a population initialization module, a fitness evaluation module, a particle velocity update module, and a particle position update module. The population initialization module generates an initial population containing multiple particles. Each particle's position vector corresponds to a candidate parameter vector, and each particle's velocity vector controls the magnitude of its position change. The fitness evaluation module uses an electromagnetic simulator to calculate the predicted power gain for each particle and evaluates the particle's performance based on these predictions. The particle velocity update module adjusts the particle's velocity vector based on its own historical best position and the population's global historical best position. The particle position update module adds the particle's current position vector to its updated velocity vector to obtain the next generation's position vector.

[0034] See Figure 3 For the current key frequency point, the current component value of the tunable element node is used as the starting point to construct a multi-dimensional parameter search space. The number of dimensions in the multi-dimensional parameter search space is equal to the sum of the total number of tunable element nodes and the total number of switching paths. Tunable element nodes include all inductor and capacitor element nodes. Each tunable element node corresponds to a continuous real number variable representing the component value; inductor element nodes correspond to inductance values, and capacitor element nodes correspond to capacitance values. Each switching path corresponds to a discrete integer variable, taking the value 0 or 1, where 0 indicates the RF switch element is in the off state and 1 indicates the RF switch element is in the on state. In the multi-dimensional parameter search space, the search range for each dimension is pre-defined based on the physical implementation constraints of the tunable element nodes and the selectable states of the switching paths. The search range for inductor element nodes is determined by the equivalent inductance value corresponding to the minimum and maximum linewidths achievable through microstrip line fabrication, and the search range for capacitor element nodes is determined by the tuning range of the varactor diode or switchable capacitor array.

[0035] In the particle swarm optimization algorithm, the number of particles in the particle population is set to 50. The position vector of each particle is represented by the symbol... The velocity vector of each particle is represented by the symbol... Representation. Position vector and velocity vector All dimensions are , This is equal to the number of dimensions in the multidimensional parameter search space. At the start of the search, the position vectors of all particles are initialized to the current element value of the tunable element node and the random perturbation value near the current on / off state code of the switch path. The velocity vectors of all particles are initialized to zero vectors or small-amplitude random vectors. The historical best position vector of each particle is represented by the symbol... The global historical best position vector of the population is represented by the symbol express.

[0036] In each iteration, a fitness evaluation operation is performed, specifically: the position vector of each particle is... The decoded values ​​are the combination of component values ​​for tunable element nodes and the on / off state codes for all switching paths, which are then input into the electromagnetic simulator. The electromagnetic simulator uses a harmonic balance simulation engine to calculate the predicted power gain. Under a set input power level, the harmonic balance simulation engine performs steady-state nonlinear analysis on the reconfigurable matching network and transistor model, solving for the fundamental and harmonic voltage and current components, and extracting the ratio of the fundamental output power to the fundamental input power as the predicted power gain. The predicted power gain is directly used as the fitness value of the particles.

[0037] After fitness evaluation, individual and global optimum updates are performed: if the current particle's fitness value is greater than its historical best fitness value, then the current particle's position vector is used to replace it. If the current particle's fitness value is greater than the population's global historical best fitness value, then replace it with the current particle's position vector. The particle velocity is updated using the following formula: in, Indicates the first The particle in the first The velocity vector at the next iteration; Indicates the first The particle in the first The velocity vector at the next iteration; Indicates the inertia weighting coefficient. The value of is linearly decreased from the initial value of 0.9 to the final value of 0.4. The linearly decreasing step size is evenly distributed according to the maximum number of iterations. The linearly decreasing inertia weight coefficient enables the particle swarm optimization algorithm to maintain a strong global exploration ability in the early stage of iteration and enhance the local fine search ability in the later stage of iteration. Indicates the first learning factor. The value of is 2. Used to control the intensity at which a particle learns to its best historical position; Indicates the second learning factor. The value of is 2. Used to control the intensity of particle learning towards the global historical best position of the population; Describes the first random vector. Each dimension component is a random number uniformly distributed in the interval [0,1]. Describes the second random vector. Each dimension component is a random number uniformly distributed in the interval [0,1]. This represents element-wise multiplication. Indicates the first The historical optimal position vector of each particle; Represents the global historical best position vector of the population; Indicates the first The particle in the first The position vector at the next iteration.

[0038] After the particle velocity update is complete, the particle position update is performed. The update formula is as follows: The updated position vector undergoes boundary processing. If a component of a dimension exceeds the upper limit of the search range for that dimension, that component is set to the upper limit of the search range, while the velocity component of that dimension is set to zero. Conversely, if a component of a dimension is below the lower limit of the search range for that dimension, that component is set to the lower limit of the search range, while the velocity component of that dimension is set to zero. The discrete component of the position vector corresponding to the switch switching path is rounded down, with values ​​less than 0.5 corrected to 0 and values ​​greater than or equal to 0.5 corrected to 1, to achieve binary encoding of the switch's on / off state.

[0039] Repeat the steps of fitness evaluation, individual optimal and global optimal update, particle velocity update, and particle position update until the change in the power gain prediction values ​​of the elite parameter vectors over 20 consecutive generations is less than a preset convergence threshold. The preset convergence threshold is set to 0.05 dB. The iterative search is considered converged when the difference between the maximum and minimum power gain prediction values ​​corresponding to the global optimal position vector of the population over 20 consecutive generations is less than 0.05 dB. The current population's global historical optimal position vector is then used. The corresponding candidate parameter vector serves as the power matching parameter set for the current key frequency point. The power matching parameter set includes the component values ​​of all tunable element nodes and the on / off state codes of all switching paths. For the implementation using a genetic algorithm as the global optimization engine, the genetic algorithm employs a chromosome structure that combines binary and real-number encoding. The selection operator uses tournament selection, the crossover operator uses simulated binary crossover, and the mutation operator uses polynomial mutation. The fitness function is also the power gain prediction value calculated by the harmonic balance simulation engine.

[0040] In practice, the power matching parameter sets corresponding to adjacent key frequencies are compared to identify parameter jump intervals where the difference exceeds a preset tolerance threshold. The key frequency sequence contains all key frequencies arranged in ascending order of frequency. Starting from the first key frequency, the first power matching parameter set for the current key frequency and the second power matching parameter set for the next key frequency are extracted sequentially. The first power matching parameter set corresponds to the lower-frequency key frequency, and the second power matching parameter set corresponds to the adjacent key frequency with a higher frequency.

[0041] The absolute difference between the component values ​​of each tunable element node at the same position in the first power matching parameter set and the second power matching parameter set is calculated to obtain the component value difference sequence. Tunable element nodes include inductor element nodes and capacitor element nodes. Each tunable element node has a first component value in the first power matching parameter set and a second component value in the second power matching parameter set. For inductor element nodes, the first and second component values ​​are the inductance values, in nanohenries; for capacitor element nodes, the first and second component values ​​are the capacitance values, in picofarads. The absolute value of the difference between the first and second component values ​​is taken as the component value difference value of the corresponding tunable element node. The component value difference values ​​of all tunable element nodes are arranged in index order to form the component value difference sequence.

[0042] This counts the number of component nodes in the statistical component value difference sequence whose absolute difference exceeds a preset component change threshold. The preset component change threshold is set according to the type of tunable component node. For inductor component nodes, the preset inductance change threshold is set to 15% of the first component value; that is, if the first inductance value is... Then the inductance change threshold is For capacitor element nodes, the preset capacitance change threshold is set to 15% of the first element value, that is, if the first capacitance value is... Then the threshold for capacitance change is When the difference in component value of a tunable element node exceeds a preset component change threshold for the corresponding type and value, the component value of that tunable element node is determined to have changed significantly, and that tunable element node is counted in the number of significantly changed component nodes. The number of significantly changed component nodes is denoted as... The total number of component nodes is denoted as The degree of difference is indicated by the symbol. Indicates the degree of difference Calculated using the following formula: in, This indicates the number of tunable element nodes in the element value difference sequence whose absolute difference exceeds a preset element change threshold. This represents the total number of tunable element nodes included in the power matching parameter set. This is a constant value that remains unchanged throughout the design process, depending on the total number of inductor and capacitor nodes in the initial topology.

[0043] The calculated difference The difference is compared with a preset tolerance threshold. The preset tolerance threshold is set to 0.3, based on the fact that when the difference exceeds 0.3, it indicates that more than 30% of the tunable element nodes have experienced significant component value jumps between two adjacent critical frequency points. This indicates a significant impedance discontinuity in the matching network within this frequency range, making simple parameter smoothing unsuitable; instead, it requires processing according to the parameter jump range. If the difference... If the frequency exceeds the preset tolerance threshold of 0.3, the frequency band between the current key frequency and the next key frequency is marked as the parameter jump interval. The lower limit frequency of the parameter jump interval is the frequency value of the current key frequency, and the upper limit frequency is the frequency value of the next key frequency. If the difference... If the value is less than or equal to the preset tolerance threshold of 0.3, it indicates that the matching parameters between adjacent key frequency points change smoothly and no parameter jump interval marker is generated.

[0044] After completing the difference comparison of all adjacent key frequency points, a set of parameter jump intervals is obtained. Each parameter jump interval corresponds to a frequency band interval and a pair of adjacent key frequency points.

[0045] The mismatched element nodes in the reconfigurable matching network are located based on parameter jump intervals. A first power matching parameter set and a second power matching parameter set corresponding to the parameter jump intervals are extracted. The element values ​​in the first power matching parameter set are compared node-by-node with those in the second power matching parameter set. The change in element value for each tunable element node is calculated as the absolute value of the difference between the second and first element values ​​divided by the absolute value of the first element value, yielding the relative change rate. All tunable element nodes are sorted from largest to smallest relative change rate, and the three tunable element nodes with the largest change rates are selected as candidate mismatched nodes. Selecting the top three as candidate mismatched nodes is based on the fact that in typical scenarios of impedance abrupt changes in the matching network, the main element nodes causing impedance jumps are usually concentrated on a few nodes; selecting three nodes achieves a balance between computational efficiency and comprehensive mismatch localization.

[0046] For each candidate mismatch node, the on / off state changes of the switching path in the matching branch containing that candidate mismatch node are tracked. Each matching branch embeds an RF switch element, and the on / off state of this RF switch element is encoded in the power matching parameter set corresponding to each key frequency point. The on / off state codes of the corresponding RF switch element in the matching branch are extracted from the first power matching parameter set corresponding to the lower limit frequency of the parameter transition interval, and from the second power matching parameter set corresponding to the upper limit frequency of the parameter transition interval. The two on / off state codes are compared for consistency. If the on / off state change record shows that the on / off state of the RF switch element in the matching branch has not changed before and after the parameter transition interval (i.e., the on / off state codes of the RF switch element in the first and second power matching parameter sets are consistent), it indicates that the matching branch has not changed its topology through switching. The drastic change in component values ​​is not caused by switch reconfiguration, but rather by the adjustment of the component values ​​of the tunable component node itself. This significant change in component values ​​is a mismatch phenomenon, and the candidate mismatch node is confirmed as a mismatched component node. If the on / off state change record shows a change in the on / off state of the RF switch element, it indicates that the change in element value is caused by switch topology reconfiguration, which is a normal network reconfiguration behavior. Therefore, this candidate mismatch node is excluded from the list of mismatched element nodes. After traversing all candidate mismatched nodes, a set of mismatched element nodes corresponding to the parameter transition intervals is obtained, which is used for subsequent switch path reconfiguration operations.

[0047] In practice, a switch path reconstruction operation is performed on the mismatched element node to generate a broadband matching parameter sequence. The current on / off state of all RF switching elements in the current matching branch where the mismatched element node is located is obtained. A mismatched element node is a tunable element node confirmed to have undergone a non-reconfigurable large-scale element value change within the parameter transition range. The matching branch where the mismatched element node is located is the current matching branch, which contains at least one RF switching element. The current on / off state is determined by the on / off state encoding of the switch switching path in the power matching parameter set corresponding to the key frequency point before the current mismatch. An on / off state encoding of 0 indicates the RF switching element is off, and an encoding of 1 indicates the RF switching element is on.

[0048] The system iterates through the backup switch paths in the current matching branch that contain backup matching element nodes, querying the alternative component value combination corresponding to each backup switch path. A backup switch path is a path formed when an RF switch element in the current matching branch is in the ON state and connected to a backup matching element node. Each backup matching element node is pre-equipped with multiple backup inductors or capacitors with different nominal values; each backup matching element node corresponds to the inductance value of a backup inductor or the capacitance value of a backup capacitor. When the current matching branch contains multiple RF switch elements, each RF switch element can be independently ON or OFF, forming multiple alternative switch path combinations. Each backup switch path corresponds to an ON / OFF combination code for an RF switch element and a set of alternative component value combinations. These alternative component value combinations are composed of the nominal component values ​​of the backup matching element nodes connected to the matching network under that switch combination and the component values ​​of the original tunable element nodes still retained in the circuit. The alternative component value combinations are pre-stored in the topology configuration table of the reconfigurable matching network. The topology configuration table is indexed by switch ON / OFF combinations and stores the connection relationships and component values ​​of each branch element.

[0049] By substituting the alternative component values ​​for each backup switch path into the electromagnetic simulator, the power gain is calculated at multiple test frequencies within the parameter transition range to obtain the bandwidth gain flatness index for each backup switch path. The parameter transition range has a lower limit frequency. and upper limit frequency ,in, The frequency value of the current key frequency point corresponding to the parameter jump interval. This represents the frequency value of the next critical frequency point corresponding to the parameter jump interval. It is generated as an arithmetic sequence within the parameter jump interval. Each test frequency point The value is 10. The arithmetic sequence is generated as follows: [The sequence is incomplete and requires further context to be fully translated.] test frequency points Determined by the following formula, ,in, From 1 to The integer sequence number.

[0050] At each test frequency point The replacement component value combinations and switch on / off combination codes are combined with the component values ​​of other unmodified matching branches to form a complete reconfigurable matching network parameter configuration, which is then input into the electromagnetic simulator. The electromagnetic simulator uses a harmonic balance simulation engine, taking the nominal input power of the power amplifier as the excitation, to calculate the output power gain, obtaining the power gain value corresponding to each test frequency point, in decibels. The test frequencies are then... The power gain value on is denoted as Band gain flatness index of the backup switch path The difference between the maximum and minimum power gain at each test frequency point within the parameter transition interval is calculated using the following formula: in, Indicates in Power gain value at each test frequency point Take the maximum value from the middle. Indicates in Power gain value at each test frequency point Take the minimum value from the middle. The unit is decibel. The smaller the value, the smaller the power gain change and the higher the gain flatness of the backup switch path within the parameter transition range.

[0051] Select the frequency band gain flatness index The backup switch path with the lowest minimum value is selected as the target reconstruction path. If multiple backup switch paths have the same minimum bandwidth gain flatness, their average insertion loss is compared, and the backup switch path with the lowest average insertion loss is selected as the target reconstruction path. The RF switches in the target reconstruction path are set to the ON state, while the RF switches in the original matching branch are set to the OFF state. The RF switches in the original matching branch are those that were ON and provided a signal path before the switch. If there are no ON RF switches in the original matching branch (i.e., the signal flows through the original inductor and capacitor nodes by default), the path switching is completed by ON of the RF switches corresponding to the target reconstruction path.

[0052] The component values ​​of the mismatched component nodes and adjacent tunable component nodes are updated to generate a new power matching parameter set covering the parameter transition range. The component values ​​of the mismatched component nodes are replaced by the component values ​​of the backup matching component nodes connected in the target reconstruction path. Adjacent tunable component nodes refer to tunable component nodes in the same matching branch or directly coupled matching branch as the mismatched component node. The component values ​​of adjacent tunable component nodes are fine-tuned through constraint optimization to keep the input reflection coefficient within the preset VSWR range within the parameter transition range. The optimized and adjusted component values ​​of the mismatched component nodes, the component values ​​of adjacent tunable component nodes, and the on / off state codes of the target reconstruction path are combined to form a new power matching parameter set. The power matching parameter sets corresponding to each key frequency point are concatenated with the new power matching parameter set in frequency order to obtain a broadband matching parameter sequence. During the concatenation process, the power matching parameter sets corresponding to the original key frequencies within the parameter transition range are removed, and the new power matching parameter sets are inserted. The broadband matching parameter sequence is arranged in order from low-frequency key frequencies to high-frequency key frequencies.

[0053] In practice, after locating the mismatched element nodes in the reconfigurable matching network based on the parameter jump intervals, the following steps are also performed. (See attached document.) Figure 4 The first operating frequency response curve of the mismatched element node before the parameter transition interval and the second operating frequency response curve after the parameter transition interval are extracted. The first operating frequency response curve refers to the power gain as a function of frequency obtained by frequency sweeping through an electromagnetic simulator within a narrow frequency range including the lower limit frequency of the parameter transition interval when the reconfigurable matched network is operating at the upper limit frequency of the parameter transition interval. The second operating frequency response curve refers to the power gain as a function of frequency obtained by frequency sweeping through an electromagnetic simulator within a narrow frequency range including the upper limit frequency of the parameter transition interval when the reconfigurable matched network is operating at the upper limit frequency. The frequency sweep range covers a frequency range from 5% below the lower limit frequency to 5% above the upper limit frequency, and the frequency sweep step size is one-thousandth of the frequency range span.

[0054] Calculate the amplitude difference curve between the first operating frequency response curve and the second operating frequency response curve. Assume the first operating frequency response curve is at a frequency... The amplitude value at that point The unit is decibels. Let the second operating frequency response curve be at frequency... The amplitude value at that point The unit is decibels, and the amplitude difference curve is indicated by the symbol. The formula for calculating the amplitude difference curve is as follows: in, This represents the operating frequency response amplitude value of the mismatched component node under the configuration corresponding to the lower limit frequency of the parameter transition range. This represents the operating frequency response amplitude value of the mismatched component node under the configuration corresponding to the upper frequency limit of the parameter transition range. This represents the frequency variable within the sweep frequency range.

[0055] The frequency points where the amplitude difference exceeds a preset resonance offset threshold are identified from the amplitude difference curve as the resonance drift frequencies. The preset resonance offset threshold is set to 3dB. The rationale is that when the gain change caused by variations in the matching network component values ​​exceeds 3dB at a certain frequency, it indicates a significant deterioration or improvement in the impedance matching at that frequency, resulting in a identifiable drift at the corresponding resonant frequency. The amplitude difference curve... Iterate through all swept frequency points and search for those that meet the requirements. dB or The frequency points in dB are marked as resonant drift frequencies.

[0056] The direction of parasitic parameter compensation for mismatched component nodes is determined based on the resonant drift frequency. This direction indicates whether to increase or decrease the parallel compensation capacitance of the mismatched component node. If the frequency of the resonant drift frequency is higher than the center frequency of the parameter jump interval, it indicates that the resonant frequency is drifting towards higher frequencies, and the equivalent parallel capacitance of the mismatched component node is too small. Therefore, the parallel compensation capacitance of the mismatched component node needs to be increased to pull the resonant frequency back to the lower frequency range. Conversely, if the frequency of the resonant drift frequency is lower than the center frequency of the parameter jump interval, it indicates that the resonant frequency is drifting towards lower frequencies, and the equivalent parallel capacitance of the mismatched component node is too large. Therefore, the parallel compensation capacitance of the mismatched component node needs to be decreased to push the resonant frequency back to the higher frequency range.

[0057] After determining the parasitic parameter compensation direction, a switch path reconstruction operation is performed on the mismatched component node. Matching compensation capacitors are selected from a pre-built compensation component library based on the parasitic parameter compensation direction. The compensation component library is a list of multiple compensation capacitors with different capacitance values ​​arranged in ascending order of capacitance, ranging from 0.1 picofarads to 10 picofarads, with a value increment of 0.1 picofarads. When selecting matching compensation capacitors, if the parasitic parameter compensation direction is to increase the parallel compensation capacitor value, the compensation capacitor with the capacitance value closest to the absolute value of the resonant frequency drift is selected from the compensation capacitors with capacitance values ​​greater than zero. The resonant frequency drift is calculated by converting the difference between the resonant drift frequency and the center frequency of the parameter jump interval using the inverse square root relationship between the resonant frequency and the capacitance value. If the parasitic parameter compensation direction is to decrease the parallel compensation capacitor value, the compensation capacitor with the best matching capacitance value is selected from the compensation capacitors with capacitance values ​​less than the current parallel capacitance value of the mismatched component node.

[0058] A compensation capacitor is connected in parallel to both ends of the mismatched element node via an additional RF switch element, forming a compensation branch. The additional RF switch element is an independent switching device separate from the existing RF switch element in the original matching branch. The first end of the additional RF switch element is connected to one end of the mismatched element node, the second end of the additional RF switch element is connected to the first end of the compensation capacitor element, and the second end of the compensation capacitor element is connected to the other end of the mismatched element node, thus forming a parallel compensation structure. The additional RF switch element uses the same model of switching device as other RF switch elements in the reconfigurable matching network to ensure consistent switching characteristics.

[0059] The conduction time sequence of the additional RF switching element in the compensation branch is adjusted so that the compensation capacitor element is connected within the frequency range corresponding to the parameter transition interval and disconnected within the frequency range outside the parameter transition interval. The conduction time sequence is determined based on the correspondence between frequency and time. In application scenarios with continuous frequency changes or transitions, the lower and upper limits of the parameter transition interval are mapped to the start and end times on the time axis. The additional RF switching element remains in the on state during the time interval from the start time to the end time and remains in the off state outside the time interval.

[0060] The access status code of the compensation branch is added to the power matching parameter set to update the equivalent component value of the mismatched component node. The access status code of the compensation branch is a binary code bit, where 1 indicates that the additional RF switch element is turned on and the compensation capacitor element is connected, and 0 indicates that the additional RF switch element is turned off and the compensation capacitor element is disconnected. The access status code of the compensation branch is added to the power matching parameter set of the frequency band where the mismatched component node is located, and the equivalent component value of the mismatched component node is updated to the equivalent capacitance value after the original component value of the mismatched component node is connected in parallel with the compensation capacitor element. The updated equivalent capacitance value of the mismatched component node is calculated using the capacitor parallel formula, that is, the equivalent capacitance value is equal to the sum of the original capacitance value and the compensation capacitance value.

[0061] After generating the broadband matching parameter sequence, the following steps are performed: Each power matching parameter set in the broadband matching parameter sequence is arranged in frequency order according to the key frequency points, forming a parameter configuration timeline. The broadband matching parameter sequence contains the power matching parameter sets corresponding to all key frequency points, as well as the new power matching parameter sets generated through the switch path reconstruction operation. All power matching parameter sets are arranged in ascending order of their corresponding key frequency point frequency values. The parameter configuration timeline uses time as the horizontal axis, with each time point corresponding to a power matching parameter set. The time points on the timeline correspond one-to-one with the key frequency points, and the frequency is mapped to time using the frequency scan rate or frequency switching schedule.

[0062] Identify the switching point between two adjacent power matching parameter sets on the parameter configuration timeline. Two adjacent power matching parameter sets correspond to two adjacent key frequency points or reconfiguration frequency points, and the switching point between two power matching parameter sets is defined as the boundary point between the end time of the previous frequency configuration and the start time of the next frequency configuration.

[0063] A switching protection time window is inserted at each switching point. The length of the switching protection time window is equal to the sum of the maximum operating time of the RF switching elements in the reconfigurable matching network and the preset stabilization margin time. The maximum operating time of the RF switching elements in the reconfigurable matching network refers to the maximum value among the times required for all used RF switching elements to switch from the off state to the on state and from the on state to the off state. The maximum operating time of the RF switching elements is obtained by consulting the technical specifications of the RF switching elements or through testing. The preset stabilization margin time is set to 500 nanoseconds. The basis for this setting is that after the RF switching element completes its operation, the internal charge migration and dielectric polarization processes of the switch require additional time to reach a steady state. A stabilization margin time of 500 nanoseconds ensures that the signal path is completely stable after the switch is switched. On the time axis at each switching point, half the length of the switching protection time window is extended forward and half the length of the switching protection time window is extended backward, forming a holding interval in which the RF switching elements do not switch. Within this holding interval, the on / off state of all RF switching elements remains unchanged.

[0064] The parameter configuration timeline after the insertion of the switching protection time window is compiled into an executable sequence of switch drive instructions. This sequence is used to sequentially control the on / off states of each RF switch element in the reconfigurable matching network. The compilation process involves traversing each stable interval on the parameter configuration timeline, extracting the on / off state that each RF switch element should maintain within that interval, and generating the corresponding level control signal for that time interval. The level control signal is a digital signal; a high level corresponds to the on state, and a low level corresponds to the off state. The level control signals of all RF switch elements within each time interval constitute a switch drive instruction. The switch drive instructions for all time intervals are arranged in chronological order to form a complete switch drive instruction sequence. This sequence of switch drive instructions is sent to the RF switch drive circuit of the reconfigurable matching network via a digital interface. The drive circuit then controls the operation of each RF switch element according to the instruction timing.

[0065] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network, characterized in that, Includes the following steps: Obtain the transistor model parameters of the power amplifier and the preset target operating frequency band range; An initial topology of the reconfigurable matching network is constructed based on the transistor model parameters. The initial topology includes multiple tunable element nodes and multiple switch-switching paths. Discretized frequency sampling processing is performed on the target operating frequency band to obtain multiple key frequency points; For each key frequency point, the component values ​​of the tunable element nodes and the on / off state of the switch switching path are dynamically adjusted through an iterative search algorithm to generate a set of power matching parameters corresponding to each key frequency point. The power matching parameter sets corresponding to adjacent key frequency points are compared to identify parameter jump intervals where the difference exceeds the preset tolerance threshold. Based on the parameter jump interval, the mismatched element node in the reconfigurable matching network is located, and a switch path reconstruction operation is performed on the mismatched element node to generate a broadband matching parameter sequence; For each key frequency point, the component values ​​of the tunable element nodes and the on / off state of the switching path are dynamically adjusted through an iterative search algorithm to generate a power matching parameter set corresponding to each key frequency point, including: For the current key frequency point, the current element value of the tunable element node is used as the search starting point to construct a multi-dimensional parameter search space; Multiple candidate parameter vectors are randomly generated within the multidimensional parameter search space. Each candidate parameter vector contains a combination of element values ​​for all tunable element nodes and the on / off state encoding of all switch switching paths. Each candidate parameter vector is input into the electromagnetic simulator to calculate the power gain, and the predicted power gain value corresponding to each candidate parameter vector is obtained. The candidate parameter vector with the highest predicted power gain is selected as the elite parameter vector, and the search neighborhood is redefined with the elite parameter vector as the center. A new generation of candidate parameter vectors is generated in the search neighborhood. Repeat the steps of power gain calculation, elite screening and search neighborhood narrowing until the change between the power gain prediction values ​​of multiple generations of elite parameter vectors is less than the preset convergence threshold, and use the current elite parameter vector as the power matching parameter set corresponding to the current key frequency point. The step of performing difference comparison processing on the power matching parameter sets corresponding to adjacent key frequency points to identify parameter jump intervals where the difference exceeds a preset tolerance threshold includes: Starting from the first key frequency point, extract the first power matching parameter set of the current key frequency point and the second power matching parameter set of the next key frequency point in sequence; Calculate the absolute difference between the component values ​​of each tunable element node at the same position in the first power matching parameter set and the second power matching parameter set to obtain the component value difference sequence; The number of component nodes whose absolute difference exceeds a preset component change threshold in the component value difference sequence is counted, and the ratio of the number of component nodes to the total number of component nodes is taken as the difference degree. The difference is compared with the preset tolerance threshold. If the difference is greater than the preset tolerance threshold, the frequency band between the current key frequency point and the next key frequency point is marked as the parameter jump interval. Locating mismatched element nodes in the reconfigurable matching network based on the parameter jump interval includes: extracting a first power matching parameter set and a second power matching parameter set corresponding to the parameter jump interval; The component values ​​in the first power matching parameter set are compared with the component values ​​in the second power matching parameter set node by node, and the top few tunable component nodes with the largest changes in component values ​​are selected as candidate mismatch nodes. For each candidate mismatch node, track the on / off state changes of the switch switching path in the matching branch where the candidate mismatch node is located; If the on / off state change record shows that the on / off state of the RF switch element in the matching branch has not changed before and after the parameter jump range, then the candidate mismatch node is confirmed as the mismatch element node. Perform a switching path reconstruction operation on the mismatched element node to generate a broadband matching parameter sequence, including: Obtain the current on / off state of all RF switching elements in the current matching branch where the mismatched element node is located; Traverse the backup switch paths that contain backup matching element nodes in the current matching branch, and query the alternative element value combination corresponding to each backup switch path; The alternative component values ​​for each backup switch path are substituted into the electromagnetic simulator, and the power gain is calculated at multiple test frequency points within the parameter jump range to obtain the frequency band gain flatness index for each backup switch path. Select the backup switching path with the best frequency band gain flatness index as the target reconstruction path, set the RF switching element in the target reconstruction path to the on state, and set the RF switching element in the original matching branch to the off state. The component values ​​of the mismatched component nodes and the component values ​​of adjacent tunable component nodes are updated to generate a new power matching parameter set covering the parameter jump range. The power matching parameter sets corresponding to each key frequency point are concatenated with the new power matching parameter set in frequency order to obtain the broadband matching parameter sequence.

2. The design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network according to claim 1, characterized in that, Constructing the initial topology of the reconfigurable matching network based on the transistor model parameters includes: The intrinsic impedance characteristics of the transistor are obtained by analyzing the parasitic capacitance and parasitic inductance values ​​in the transistor model parameters. The target impedance transformation values ​​for the input and output ports of the matching network are determined based on the intrinsic impedance characteristics of the transistor. Select multiple inductor node and multiple capacitor node, and connect the inductor node and the capacitor node in series or parallel to form multiple matching branches; An RF switch element is embedded in each of the matching branches as the switching path. The first end of the RF switch element is connected to the input end of the inductor node or the capacitor node, and the second end of the RF switch element is connected to the spare matching node.

3. The design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network according to claim 2, characterized in that, Discretized frequency sampling is performed on the target operating frequency band to obtain multiple key frequency points, including: Obtain the lower and upper frequency values ​​of the target operating frequency band range; The frequency band between the lower limit frequency value and the upper limit frequency value is divided into equal intervals according to the logarithmic frequency scale to obtain multiple candidate frequency points; For each candidate frequency point, calculate the frequency ratio between that candidate frequency point and its adjacent candidate frequency points; Candidate frequency points whose frequency ratio is less than a preset uniformity threshold are identified as redundant frequency points and removed, while the remaining candidate frequency points are retained as the key frequency points.

4. The design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network according to claim 1, characterized in that, The iterative search algorithm uses particle swarm optimization or genetic algorithm as the global optimization engine.

5. The design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network according to claim 1, characterized in that, The electromagnetic simulator uses a harmonic balance simulation engine to calculate the predicted power gain value.

6. The design method for a broadband microwave solid-state power amplifier based on a reconfigurable matching network according to claim 1, characterized in that, The frequency band gain flatness index is obtained by calculating the difference between the maximum and minimum power gain at each test frequency point within the parameter jump interval.

Images