Research On Large Signal Statistical Model Of GaN HEMT Microwave Device

With the rapid development of wireless communication,radar and electronic warfare and other fields,people put forward higher requirements on the performance of the microwave transceiver.As the key component of the transmitter,power amplifier,the most important and the most critical power transistor,its bandwidth,Pout,PAE and operating temperature and other performance will seriously affect the whole performance of the system.Compared with Si,the representative of the first generation of semiconductor,and GaAs,the representative of the second generation of semiconductor materials,the third generation of semiconductor materials,GaN material has the advantages of large band gap,high electron saturation velocity,high breakdown electric field and high thermal conductivity,so it becomes a important research topic in semiconductor devices.The model of GaN plays a key role in circuit design,the applicability and accuracy of the model are very important to the design of microwave circuits.At present,some progress has been made in the physical model,small signal and large signal equivalent model of GaN HEMT.However,due to the production process of the process error,model uncertainty,the change of process parameters and environmental factors,the analysis and optimization design of the device' s and the circuit' s consistency and process analysis are very important.So it is very important to establish the statistical model of GaN microwave devices.In this paper,the GaN HEMT equivalent circuit model is studied,the statistical model of large and small signal equivalent circuit is established;Based on the equivalent circuit model,the HEMT GaN high efficiency power amplifier is designed to verify the statistical model.The work carried out and its innovations are as follows:1.The statistical model of GaN HEMT small signal equivalent circuit based on Monte Carlo methodPoint at the production process of process error caused by the equivalent circuit model's accuracy,test the S parameters of 56 GaN HEMT devices from 10 batches under different bias,the parameter extraction is carried out.A GaN HEMT statistical model under different bias is presented to obtain the equivalent circuit parameters of the small signal.The modeling methods include principal component analysis,factor analysis,multiple regression model and Monte Carlo method.When applied to theequivalent circuit parameters of GaN HEMT devices,by comparing the original and the model of the mean,standard deviation,correlation matrix and S parameters,and analyzing the S parameters under different biases,the precision of the small signal equivalent circuit statistical model is proved.2.The statistical model of GaN HEMT's large signal equivalent circuit based on Monte Carlo methodPoint at the volatility of large signal indicators(output power Pout,power added efficiency PAE,etc.),test the DC IV and large signal results of 34 GaN HEMT devices from 10 batches,the parameters of Ids,Cgs and Cgd expressions of nonlinear source and drain source are extracted.Based on small signal equivalent circuit modeling statistical model,a whole large-signal equivalent circuit parameters of statistical models is presented to overcome the shortcoming that single model can not reflect the fluctuation range of the actual circuit.The obtained large signal statistical model is embedded in ADS to carry out simulation,by comparing the simulation results with the measured drain-source current,output power and power added efficiency,the statistical properties of the output power and power added efficiency are analyzed by means of simulation and test,the precision of the large signal statistical model established by the above method is proved.3.The statistical model of GaN HEMT's large signal equivalent circuit based on response surface methodPoint at the large amount of data in statistical models leading to convergence problems in circuit simulation.Based on response surface method,and the response surface method is optimized and improved,presents a simple large signal statistical model.By using the response surface method,the statistical model has the advantages of small amount of data,fast,simple and accurate,and the statistical model has a good convergence in the circuit simulation.By comparing the simulation results with the measured drain-source current,output power and power added efficiency,the statistical properties of the output power and power added efficiency are analyzed by means of simulation and test,the precision of the large signal statistical model established by the response surface method is proved.4.Design and analysis of HEMT GaN high efficiency power amplifier based on statistical modelIn order to verify the accuracy of the established statistical models,based on theestablished GaN HEMT equivalent circuit statistical model from Monte Carlo method and the response surface method,three E class 13.7GHz~14.2GHz power amplifiers with Ku-band working band are designed and three inverse F class 2.7GHz~3.5GHz power amplifiers with S-band working band are designed and manufactured.The comparison between simulation and test results show that the measured results are in the range of simulation results,and the mean value is very close to the simulation results.It is proved that the statistical model established above is very accurate and suitable for circuit design and yield analysis.The application of statistical model in a C band 60 W GaN HEMT power amplifier circuit MMIC for simulation,by comparing with the measured results,further verify the accuracy of the statistical model of large signal,at the same time to illustrate the statistical model is suitable for the power amplifier circuit design and yield analysis.