Modeling of RF power transistors for power amplifier design.

摘要

Presented is a large/small signal, non-quasi-static, charge conserving, MOSFET modeling technique suitable for DC and high frequency circuit design. The technique is theoretically developed for a SOI MOSFET on a physical device simulator and is applied to a LDMOSFET device.; A DC electro-thermal model is developed by using a single thermal resistance to map the thermal response of the device. An automated novel approach to measuring iso-thermal IVs and microwave scattering parameters is discussed as an alternative to pulsed IV, pulsed RF measurements. The iso-thermal IV is used by the device in establishing its DC point of operation. The iso-thermal s-parameters are used to extract temperature dependent small signal model parameters. A novel extraction technique utilizing analytical expressions is presented. The extraction procedure yields a continuum of solutions as a function of the source parasitic resistances, Rs. A multi-bias analysis is used to determine the final Rs. gm, C11, and C 21 are found to vary with temperature while gd, C12, and C22 are found to exhibit much less temperature variation.; Hot and cold pulsed IVs are compared with iso-thermal IVs. Their associated transconductances are compared to RF gm, obtained from microwave measurements. The existence of low frequency dispersion in LDMOSFETs, in spite of the p+ sinker, is demonstrated. It is verified that globally over temperature, the iso-thermal IVs still agree the best with gm,RF.; A temperature dependent large signal model is generated by integrating temperature dependent gm,RF and g d,RF. 3D Tensor Product B-Splines (TPS) utilizing an optimum knot placement scheme, are used to represent the IV and it's associate derivatives. Optimum knot placement is necessary to handle areas of large derivative variation in data. Optimized TPS is used to represent the capacitances and extract the gate and drain charges.; The large signal model is implemented in ADS as a user defined model. In the process of doing so, a number of bugs in ADS were reported to Agilent. Harmonic balance simulations are conducted on a power amplifier and amplifier matrices such as fundamental, second and third harmonic power, drain current, temperature and two tone intermodulation distortion predicted by the model are compared to measured results. The impact on the model of temperature dependent drain and gate charge is investigated. Higher temperature is found to introduce more profound sweet spots. The model, in spite of its simplifications, is found to compare well with the existing MET model, which is based on pulsed IV and pulsed RF. For two tone IMD simulations, the developed model is found to out perform the MET model.