CN106126804A - The behavioral scaling modeling of a kind of power amplifier bottom circuit and verification method - Google Patents

Abstract

The present invention relates to behavioral scaling modeling and the verification method of a kind of power amplifier bottom circuit, comprise: S1, set up power amplifier bottom circuit model, gather the input signal of power amplifier and output signal as experimental data；S2, foundation behavior model based on mathematic(al) representation, utilize experimental data that behavior model is carried out identification；S3, build identical input signal exciting circuit, calculate the normalized mean squared error between power amplifier bottom circuit model and behavior model output signal；S4, judge whether normalized mean squared error meets required precision, verify the behavior model accurate description to power amplifier bottom circuit input-output characteristic；As being unsatisfactory for, readjust behavior model and meet required precision to normalized mean squared error.The present invention solves power amplifier bottom circuit Straight simulation and solves a difficult problem for complexity, can apply to the not exclusively known simulation scenarios of bottom circuit details caused due to technical know-how, the System Level Electromagnetic Compatibility analysis for analog-and digital-hybrid circuit provides effective means.

Description

The behavioral scaling modeling of a kind of power amplifier bottom circuit and verification method

Technical field

The present invention relates to behavioral scaling modeling and the verification method of a kind of power amplifier bottom circuit, specifically refer to a kind of use based on The behavior model of mathematic(al) representation describes modeling and the verification method of the input-output characteristic of power amplifier bottom circuit, solves at the bottom of power amplifier A complicated calculations difficult problem for a large amount of integro-differential equations in layer circuit model nonlinear characteristic solution procedure, belongs to electromagnetic compatibility technology neck Territory.

Background technology

EMC analysis is in addition to the normal function of simulation computing system, in addition it is also necessary to carry out spectral artifacts and behavior Analyze, thus prediction and diagnosis interference phenomenon.The nonlinear characteristic of communication system intermediate power amplifier is to produce electromagnetic compatibility to ask The main cause of topic, in order to improve efficiency, power amplifier needs to be operated near even saturation region, inelastic region；But non-linear phenomena Substantially, spectral re-growth will be inevitably resulted in, thus adjacent channel or signal will be interfered, and the most also can affect self function Normal realization.

Generally, the modeling and simulating of power amplifier is to calculate by setting up bottom circuit, analog-digital hybrid circuit system The solution procedure of system will also tend to substantial amounts of non-linear long-pending peaker equation so that solve difficulty；Meanwhile, power amplifier bottom The details of circuit are more difficult acquisition in the case of technical know-how；Accordingly, it would be desirable to use higher by one than transistor device structures The emulation technology of level, it is only necessary to set up the input-output characteristic model of power amplifier.

Behavioral scaling modeling technique is undoubtedly the important means of system-level EMC analysis, describes each subsystem by setting up System and the functional mode of circuit module, the simplification to underlying physical structure circuit will be greatly improved simulation velocity.2012 open The Chinese invention patent of Application No. 201210187626.9, it is provided that a kind of based on the electronic devices and components Electromagnetic Launching measured Broadband behavioral scaling prediction modeling method, it establishes the behavior model of electronic circuit by measurement data, for SPICE (Simulation Program with Integrated Circuit Emphasis) circuit simulation.And disclosed in 2009 The Chinese invention patent of Application No. 200910238424.0, it is provided that a kind of circuit board level electromagnetic compatible sensitivity behavioral scaling is built Modular system, it is by setting up IBIS (the Input/Output Buffer Informational of electronic circuit Specification) behavior model, is applied to whole circuit board level comprehensive simulating.The emulation of these behavioral scalings is main by setting up It is analyzed with the port model of bottom circuit compatibility, it is adaptable to relatively ball bearing made and the situation of device, for complicated power amplifier For circuit, the behavior model of foundation is the most complicated.

Based on above-mentioned, need behavioral scaling modeling and the checking proposing a kind of power amplifier bottom circuit based on mathematical model at present badly Method, solution power amplifier bottom circuit details are the most known and emulation solves a complicated difficult problem, for system-level electromagnetic compatibility Property analysis provide effective means.

Summary of the invention

It is an object of the invention to provide behavioral scaling modeling and the verification method of a kind of power amplifier bottom circuit, solve power amplifier Bottom circuit Straight simulation solves a difficult problem for complexity, and the bottom circuit that simultaneously can also be applied to cause due to technical know-how is detailed Simulation scenarios known to INFORMATION OF INCOMPLETE, the System Level Electromagnetic Compatibility analysis for analog-and digital-hybrid circuit provides effectively Means.

In order to achieve the above object, the present invention provides behavioral scaling modeling and verification method, the bag of a kind of power amplifier bottom circuit Containing following steps:

S1, set up power amplifier bottom circuit model, gather the input signal of power amplifier and output signal as experimental data；

S2, set up behavior model based on mathematic(al) representation, utilize the input signal collected in S1 and output signal Experimental data carries out identification to behavior model；

S3, build identical input signal exciting circuit, calculate power amplifier bottom circuit model and behavior model output signal Between normalized mean squared error；

S4, judge whether normalized mean squared error meets required precision, the accuracy of checking behavior model；As met, complete Become behavioral scaling modeling and the checking of power amplifier bottom circuit；As being unsatisfactory for, then return S2, readjust behavior model, until normalizing Change mean square error and meet required precision, it is achieved the behavior model accurate description to power amplifier bottom circuit input-output characteristic.

In described S1, power amplifier bottom circuit model is the circuit using power discharging device, resistance, electric capacity or microstrip line to build Model.

In described S1, the input signal of power amplifier and output signal are the baseband signal of discrete form, useRepresent defeated Enter signal,Represent output signal.

In described S2, behavior model based on mathematic(al) representation refers to describe the mathematical expression of power amplifier input-output characteristic Formula, physical relationship formula is:

$\tilde{y}\left(l\right)=f\left(\tilde{x}\right(l\left)\right);$

Wherein, f () function is Taylor series, or memory polynomial function, or Volterra progression.

In described S2, behavior model is carried out identification and refers to utilize the input signal collected in S1 and output signal Experimental data solve the coefficient to be estimated of f () function；

When the experimental data of the input signal collected in S1 and output signal is more than the number of coefficient to be estimated, use Method of least square carries out identification, estimates that the relational expression of the coefficient of f () function is:

$\hat{H}={\left(X^{\′}X\right)}^{-1}{X}^{\′}Y;$

When the experimental data of the input signal collected in S1 and output signal is less than the number of coefficient to be estimated, use Minimum L2 norm method carries out identification, estimates that the relational expression of the coefficient of f () function is:

$\hat{H}=X^{\′}{\left({\mathrm{XX}}^{\′}\right)}^{-1}Y;$

Wherein,The coefficient matrix of the f () function for estimating, X is the input signal matrix set up, and X' is the conjugation of X Transposed matrix, Y is the output signal matrix set up.

In described S3, identical input signal exciting circuit refers to the front end of power amplifier bottom circuit model and behavior model Excitation keeps consistent.

In described S3, the normalized mean squared error between calculating power amplifier bottom circuit model and behavior model output signal Relational expression is:

$E_{NMSE}=10log\left(\frac{\underset{l=1}{\overset{L}{\Σ}}\left(\right|{\tilde{y}}_l-{\tilde{y}}_l^{\mathrm{mod}}{|}^2)}{\underset{l=1}{\overset{L}{\Σ}}|{\tilde{y}}_l{|}^2}\right);$

Wherein, L is sampled point number,For the output signal of behavior model,Experimental data for output signal.

In sum, the behavioral scaling modeling of the power amplifier bottom circuit that the present invention provides and verification method, use based on mathematics The behavior model of expression formula describes the input-output characteristic of power amplifier bottom circuit, solves power amplifier bottom circuit Straight simulation and solves A complicated difficult problem, can also be applied to the not exclusively known emulation of bottom circuit details caused due to technical know-how simultaneously Situation, by utilizing behavior model comprehensive and Substitute For Partial bottom circuit, and behavior model can also be the most double with circuit model Holding, the System Level Electromagnetic Compatibility analysis for analog-and digital-hybrid circuit provides effective means.

Accompanying drawing explanation

Fig. 1 is the behavioral scaling modeling flow chart with verification method of the power amplifier bottom circuit in the present invention；

Fig. 2 is the structural representation of the specific embodiment of the power amplifier bottom circuit model in the present invention；

Fig. 3 is the structural representation of the specific embodiment of the input signal exciting circuit in the present invention.

Detailed description of the invention

Below in conjunction with Fig. 1～Fig. 3, describe a preferred embodiment of the present invention in detail.

As it is shown in figure 1, the behavioral scaling for the power amplifier bottom circuit of present invention offer models and verification method, comprise following step Rapid:

S1, set up power amplifier bottom circuit model, gather the input signal of power amplifier and output signal as experimental data；

S2, set up behavior model based on mathematic(al) representation, utilize the input signal collected in S1 and output signal Experimental data carries out identification to behavior model；

S3, build identical input signal exciting circuit, calculate power amplifier bottom circuit model and behavior model output signal Between normalized mean squared error；

S4, judge whether normalized mean squared error meets required precision, the accuracy of checking behavior model；As met, say The behavior model accurate description of bright foundation power amplifier bottom circuit, completes behavioral scaling modeling and the checking of power amplifier bottom circuit；As It is unsatisfactory for, illustrates that the behavior model set up fails accurate description power amplifier bottom circuit, need to return S2, readjust behavior mould Type, until normalized mean squared error meets required precision, it is achieved the behavior model standard to power amplifier bottom circuit input-output characteristic Really describe.

In described S1, power amplifier bottom circuit model is to use the components and parts such as power discharging device, resistance, electric capacity or microstrip line to take The circuit model built.

In described S1, the input signal of power amplifier and output signal are the baseband signal of discrete form, useRepresent defeated Enter signal,Represent output signal.

In described S2, behavior model based on mathematic(al) representation refers to describe the mathematical expression of power amplifier input-output characteristic Formula, physical relationship formula is:

$\tilde{y}\left(l\right)=f\left(\tilde{x}\right(l\left)\right);$

Wherein, f () function is Taylor series, or memory polynomial function, or Volterra (Volterra) progression.

In described S2, behavior model is carried out identification and refers to utilize the input signal collected in S1 and output signal Experimental data solve the coefficient to be estimated of f () function；

When the experimental data of the input signal collected in S1 and output signal is more than the number of coefficient to be estimated, use Method of least square carries out identification, estimates that the relational expression of the coefficient of f () function is:

$\hat{H}={\left(X^{\′}X\right)}^{-1}{X}^{\′}Y;$

When the experimental data of the input signal collected in S1 and output signal is less than the number of coefficient to be estimated, use Minimum L2 norm method carries out identification, estimates that the relational expression of the coefficient of f () function is:

$\hat{H}=X^{\′}{\left({\mathrm{XX}}^{\′}\right)}^{-1}Y;$

Wherein,The coefficient matrix of the f () function for estimating, X is the input signal matrix set up, and X' is the conjugation of X Transposed matrix, Y is the output signal matrix set up.

In described S3, identical input signal exciting circuit refers to the front end of power amplifier bottom circuit model and behavior model Excitation keeps consistent, and simulation calculation can compatible circuit model and mathematics behavior model.

In described S3, the normalized mean squared error between calculating power amplifier bottom circuit model and behavior model output signal Relational expression is:

$E_{NMSE}=10log\left(\frac{\underset{l=1}{\overset{L}{\Σ}}\left(\right|{\tilde{y}}_l-{\tilde{y}}_l^{\mathrm{mod}}{|}^2)}{\underset{l=1}{\overset{L}{\Σ}}|{\tilde{y}}_l{|}^2}\right);$

Wherein, L is sampled point number,For the output signal of behavior model,Experimental data for output signal.When Calculated normalized mean squared error E_NMSEThe least, then illustrate that the behavior model set up is the most accurate.

Below by way of a specific embodiment of the present invention, describe the inventive method in detail in the internal merit of mobile communication system The modeling of rate amplifier behavior level circuit and the application in checking.In the present embodiment, owing to employing simulation result is as experimental data, Without the concern for test error, therefore it is required that the normalized mean squared error of behavior model is better than (being less than)-35dB, i.e. meet precision Requirement.Specifically follow the steps below the behavioral scaling modeling of power amplifier bottom circuit and verify:

S1, set up power amplifier bottom circuit model, gather the input signal of power amplifier and output signal as experimental data.

In the present embodiment, the power amplifier within certain mobile communication system be used for amplifying code check be 3.84Mcps (megacycle/ Second) WCDMA (WCDMA, Wideband Code Division Multiple Access) signal.The merit set up Put bottom circuit model as in figure 2 it is shown, be made up of power discharging device 1, multiple electric capacity 2 and multiple microstrip line 3, and by DC source 4 power supplies.Wherein, the carrier frequency of input signal is 2.14GHz, and power amplifier output is 10W.Utilize ADS simulation software pair After the power amplifier bottom circuit of WCDMA signal excitation emulates, gather 2000 input signals and output signal as experiment number According to carrying out behavior model identification.

S2, set up behavior model based on mathematic(al) representation, utilize the input signal collected in S1 and output signal Experimental data carries out identification to behavior model.

In the present embodiment, use Taylor series as f () function, it is considered to the bandpass characteristics of power amplifier is by even order terms Filter, according toThe expression formula obtaining behavior model is:

$\tilde{y}\left(l\right)=\underset{n=1}{\overset{N}{\Σ}}{a}_{2n-1}|\tilde{x}\left(l\right){|}^{2(n-1)}\tilde{x}\left(l\right);$

Wherein, a_2n-1For coefficient to be estimated；

Choosing exponent number N=5, coefficient to be estimated is less than the input signal and the experimental data number of output signal collected, UtilizeCarry out identification, obtain the coefficient matrix of the f () function estimated

S3, build identical input signal exciting circuit, calculate power amplifier bottom circuit model and behavior model output signal Between normalized mean squared error.

In the present embodiment, ADS simulation software is built WCDMA signal exciting circuit, as it is shown on figure 3, WCDMA signal warp After being generated by signal generating circuit, respectively enter the power amplifier bottom circuit model set up with ADS simulation software and use ADS- The mathematics behavior model that MATLAB Cooperative Analysis instrument is set up, and respectively by power amplifier output signal Acquisition Circuit and behavior model Output signal Acquisition Circuit gathers output signal, thus is calculated between power amplifier bottom circuit model and behavior model output signal Normalized mean squared error be-30.6dB.

S4, judge whether normalized mean squared error meets required precision, the accuracy of checking behavior model；As met, say The behavior model accurate description of bright foundation power amplifier bottom circuit；As being unsatisfactory for, illustrate that the behavior model set up fails accurately to retouch State power amplifier bottom circuit, need to return S2, readjust behavior model, until normalized mean squared error meets required precision, real The existing behavior model accurate description to power amplifier bottom circuit input-output characteristic.

In the present embodiment, calculated normalized mean squared error is-30.6dB, and it is more than-35dB, therefore not met essence Degree requirement；It is thus desirable to return S2 to readjust the form of f () function, this time use memory polynomial function as f () Function, the expression formula obtaining behavior model is:

$\tilde{y}\left(l\right)=\underset{m=0}{\overset{M-1}{\Σ}}\underset{n=1}{\overset{N}{\Σ}}{a}_{2n-1,m}|\tilde{x}(l-m){|}^{2(n-1)}\tilde{x}(l-m);$

Wherein, a_2n-1,mFor coefficient to be estimated.

Choosing exponent number N=5, memory span M=4, coefficient to be estimated is less than the input signal collected and output signal Experimental data number, utilizeCarry out identification, obtain the coefficient matrix of the f () function estimatedAnd count Calculating the normalized mean squared error that obtains between power amplifier bottom circuit model and behavior model output signal be-37.8dB, it is less than- 35dB, therefore meets required precision, completes behavioral scaling modeling and the checking of power amplifier bottom circuit.

Compared with prior art, the present invention provide power amplifier bottom circuit behavioral scaling modeling and verification method, have with Lower advantage and beneficial effect:

1, the functional character of mathematics behavior model accurate description power amplifier bottom circuit, the simplification to power amplifier bottom circuit are utilized Simulation velocity will be greatly improved, and reduce calculating resource；

2, the mathematics behavior model set up has versatility, and model parameter is easy to adjust, proposes merit the most from top to bottom The index request on electric discharge road, carries out analysis and the assessment of power amplifier nonlinear characteristic according to the actual requirements；

3, behavioral scaling emulation meets the technical know-how requirement of power amplifier design manufacturer, it is not necessary to provide the internal electricity in detail of power amplifier Road, it is only necessary to provide the behavior model of power amplifier, i.e. can be used for the emulation of whole communication circuitry level；

4, mathematics behavior model can also be compatible with bottom circuit model, simplifies the interface between phantom, improves Analysis efficiency.

Although present disclosure has been made to be discussed in detail by above preferred embodiment, but it should be appreciated that above-mentioned Description is not considered as limitation of the present invention.After those skilled in the art have read foregoing, for the present invention's Multiple amendment and replacement all will be apparent from.Therefore, protection scope of the present invention should be limited to the appended claims.

Claims (7)

1. the behavioral scaling of a power amplifier bottom circuit models and verification method, it is characterised in that comprise the steps of

S1, set up power amplifier bottom circuit model, gather the input signal of power amplifier and output signal as experimental data；

S2, foundation behavior model based on mathematic(al) representation, utilize the experiment of input signal and the output signal collected in S1 Data carry out identification to behavior model；

S3, build identical input signal exciting circuit, calculate between power amplifier bottom circuit model and behavior model output signal Normalized mean squared error；

S4, judging whether normalized mean squared error meets required precision, checking behavior model is to power amplifier bottom circuit input and output The accurate description of characteristic；As met, complete behavioral scaling modeling and the checking of power amplifier bottom circuit；As being unsatisfactory for, then return S2, weight Newly adjust behavior model, until normalized mean squared error meets required precision.

2. the behavioral scaling of power amplifier bottom circuit as claimed in claim 1 models and verification method, it is characterised in that described S1 In, power amplifier bottom circuit model is the circuit model using power discharging device, resistance, electric capacity or microstrip line to build.

3. the behavioral scaling of power amplifier bottom circuit as claimed in claim 2 models and verification method, it is characterised in that described S1 In, the input signal of power amplifier and output signal are the baseband signal of discrete form, useRepresent input signal,Represent Output signal.

4. the behavioral scaling of power amplifier bottom circuit as claimed in claim 3 models and verification method, it is characterised in that described S2 In, behavior model based on mathematic(al) representation refers to describe the mathematic(al) representation of power amplifier input-output characteristic, and physical relationship formula is:

$\tilde{y}\left(l\right)=f\left(\tilde{x}\left(l\right)\right);$

Wherein, f () function is Taylor series, or memory polynomial function, or Volterra progression.

5. the behavioral scaling of power amplifier bottom circuit as claimed in claim 4 models and verification method, it is characterised in that described S2 In, behavior model is carried out identification and refers to that the experimental data utilizing input signal and the output signal collected in S1 solves f The coefficient to be estimated of () function；

When the experimental data of the input signal collected in S1 and output signal is more than the number of coefficient to be estimated, use minimum Square law carries out identification, estimates that the relational expression of the coefficient of f () function is:

$\hat{H}={\left(X^{\′}X\right)}^{-1}{X}^{\′}Y;$

When the experimental data of the input signal collected in S1 and output signal is less than the number of coefficient to be estimated, use minimum L2 norm method carries out identification, estimates that the relational expression of the coefficient of f () function is:

$\hat{H}=X^{\′}{\left({\mathrm{XX}}^{\′}\right)}^{-1}Y;$

Wherein,The coefficient matrix of the f () function for estimating, X is the input signal matrix set up, and X' is the conjugate transpose square of X Battle array, Y is the output signal matrix set up.

6. the behavioral scaling of power amplifier bottom circuit as claimed in claim 5 models and verification method, it is characterised in that described S3 In, identical input signal exciting circuit refers to that power amplifier bottom circuit model keeps consistent with the excitation of the front end of behavior model.

7. the behavioral scaling of power amplifier bottom circuit as claimed in claim 6 models and verification method, it is characterised in that described S3 In, the relational expression calculating the normalized mean squared error between power amplifier bottom circuit model and behavior model output signal is:

$E_{NMSE}=10log\left(\frac{\underset{l=1}{\overset{L}{\Σ}}\left({\left|{\tilde{y}}_l-{\tilde{y}}_l^{\mathrm{mod}}\right|}^2\right)}{\underset{l=1}{\overset{L}{\Σ}}{\left|{\tilde{y}}_l\right|}^2}\right);$

Wherein, L is sampled point number,For the output signal of behavior model,Experimental data for output signal.