Extracting a bias-dependent large signal MESFET model from pulsedI/V measurements

In this paper a new large-signal metal semiconductor field effect transistor (MESFET) model suitable for applications to nonlinear microwave CAD has been developed and the different phenomena involved in the nonlinear behavior of the transistor have been studied. The importance of this work lies in the fact that multibias starting points (hot and cold device) for pulsed measurements are used to derive a single expression for I_ds that describes the dc as well as the small and large signal behavior of the transistor, while taking into account the quiescent point dependence. The algorithms of this new model can easily be incorporated into commercially available nonlinear simulators. The operating-point dependent current I_ds is modeled by two nonlinear sources: one of them is the dc characteristic nonlinear equation, and the other represents the differences between dc and pulsed characteristics at every bias point. A complete large-signal model is presented for a 10*140 μm GaAs-MESFET chip (F20 process) from the GEC-MARCONI Foundry and a 16*250 μm MESFET chip (DIOM process) from the Siemens Foundry. Comparisons have been made between simulations and measurements of pulsed characteristics at different operating points. There was very good agreement between the P_in /P_out measurements and the MDS simulations using the complete large signal model     

Modeling of DC–DC Buck Converters for Large-Signal Frequency Response and Limit Cycles

A zeroth-order-hold equivalent discrete-time model of the buck converter for computing its large-signal frequency response is developed and experimentally verified. It is shown that, with a dc bias and a sinusoidal variation of the input duty cycle, the frequency response of the output voltage from the converter shifts from underdamped behavior to damped behavior with increasing amplitude of the input sinusoid. It is observed that, with a given dc input bias and a given input amplitude beyond the range of the state-space linearized small-signal model, the converter behavior varies from exclusively continuous inductor current mode at low frequencies to behavior with continuous and discontinuous inductor current modes at high frequencies. The use of this sinusoidal input large-signal frequency response in predicting limit cycles induced by feedback of the output voltage using proportional and integral controllers for such converters is studied. Experimental results confirming the use of this large-signal frequency response are presented.     

An analysis of small-signal and large-signal base resistances forsubmicrometer BJT's

Small-signal and large-signal base resistances for submicrometer BJT's have been analyzed in detail by using a two-dimensional device simulator. It has been clarified that the small-signal base resistance cannot be obtained from the input impedance semicircle in the complex plane accurately due to the effect of the emitter resistance at low frequencies, and also due to the reduction of the base-emitter junction resistance and the diffusion capacitance at high frequencies. A new method for extracting an accurate value of the small-signal base resistance is proposed. The extracted small-signal base resistance not only differed from the large-signal base resistance extracted from the dc characteristics of a BJT, but also agreed with the differential resistance. It has been shown that, for a 0.6 μm emitter-width BJT, the measured value of the small-signal base resistance is less than half that of the large-signal base resistance in the bias region in which actual BJT circuits operate     

Accurate pHEMT nonlinear modeling in the presence of low-frequency dispersive effects

Low-frequency (LF) dispersive phenomena due to device self-heating and/or the presence of "traps" (i.e., surface state densities and bulk spurious energy levels) must be taken into account in the large-signal dynamic modeling of III-V field-effect transistors when accurate performance predictions are pursued, since these effects cause important deviations between direct current (dc) and dynamic drain current characteristics. In this paper, a new model for the accurate characterization of these phenomena above their cutoff frequencies is presented, which is able to fully exploit, in the identification phase, large-signal current-voltage (I-V) measurements carried out under quasi-sinusoidal regime using a recently proposed setup. Detailed experimental results for model validation under LF small- and large-signal operating conditions are provided. Furthermore, the I-V model proposed has been embedded into a microwave large-signal pseudomorphic high electron-mobility transistor (pHEMT) model in order to point out the strong influence of LF modeling on the degree of accuracy achievable under millimeter-wave nonlinear operation. Large-signal experimental validation at microwave frequencies is provided for the model proposed, by showing the excellent intermodulation distortion (IMD) predictions obtained with different loads despite the very low power level of IMD products involved. Details on the millimeter-wave IMD measurement setup are also provided. Finally, IMD measurements and simulations on a Ka-band highly linear power amplifier, designed by Ericsson using the Triquint GaAs 0.25-/spl mu/m pHEMT process, are shown for further model validation.     

A unified model for single/multifinger HBTs including self-heatingeffects

This paper presents a unified analytical large-signal model that includes self-heating effects. The model is applied to a single-finger AlGaAs/GaAs heterojunction bipolar transistor (HBT) and a multifinger InGaAs/GaAs HBT. The self-heating effect in the HBT is simulated as a feedback from the collector current to the base-emitter voltage. The main advantage of the circuit presented here is that additional analysis of coupling between electrical and thermal circuits is not required, as is the case with the existing models. The small-signal HBT model is implemented based on the S-parameters at multiple frequencies measured at multiple bias points. This model is verified by comparing the measured and simulated S-parameters. The large-signal model is based on the forward Gummel plot and is built over the small-signal model. This model is verified by comparing the simulated and measured dc I-V characteristics     

Thermal analysis of AlGaN-GaN power HFETs

In this paper, we present a thermal analysis of AlGaN-GaN power heterojunction field-effect transistors (HFETs). We report the dc, small-signal, large-signal, and noise performances of AlGaN-GaN HFETs at high temperatures. The temperature coefficients measured for GaN HFETs are lower than that of GaAs pseudomorphic high electron-mobility transistors, confirming the potential of GaN for high-temperature applications. In addition, the impact of thermal effects on the device dc, small-signal, and large-signal characteristics is quantified using a set of pulsed and continuous wave measurement setups. Finally, a thermal model of a GaN field-effect transistor is implemented to determine design rules to optimize the heat flow and overcome self-heating. Arguments from a device, circuit, and packaging perspective are presented.     

A new empirical large-signal HEMT model

We propose an empirical large-signal model of high electron mobility transistors (HEMTs). The bias-dependent data of small-signal equivalent circuit elements are obtained from S-parameters measured at various bias settings. And C_gs, C_gd, g_m, and g_ds, are described as functions of V_gs and V_ds. We included our large-signal model in a commercially available circuit simulator as a user-defined model and designed a 30/60-GHz frequency doubler. The fabricated doubler's characteristics agreed well with the design calculations     

Nonlinear analysis of the avalanche transit-time oscillator

The nonlinear operating characteristics of the avalanche transit-time oscillator are studied by means of Fourier-series representation. For optimum operation, the oscillator must be designed such that start-oscillation conditions are satisfied simultaneously at the first and the second harmonic of the desired oscillation frequency. Under those conditions the oscillation frequency does not depend on the dc bias current; the signal level increases smoothly with bias current. For large signals, the diode exhibits negative resistance for frequencies substantially below the avalanche frequency; the oscillation frequency therefore may be below the avalanche frequency corresponding to the dc bias current required for large-signal operation. A condition for attaining large-signal operation is that the product of drift-zone capacitance and total load resistance must be small compared to the oscillation period; this condition also yields small starting currents. The output power at the oscillation frequency is obtained explicitly in terms of diode and external circuit parameters. The maximum attainable output power is limited by parasitic series resistance and by permissible RF voltage swing as compared to dc bias voltage. The best power-impedance product is obtained by choosing the transit angle equal to 0.74 π. In practice, it may be advantageous to choose a smaller value for the transit angle, in order that the tuning condition for the second harmonic may be more easily satisfied. The dc-to-RF conversion efficiency in principle is linearly proportional to the dc current density; the maximum efficiency again is limited by parasitic series resistance and by permissible RF voltage swing.     

Power performance and scalability of AlGaN/GaN power MODFETs

The scalability of power performance of AlGaN/GaN MODFETs with large gate periphery, as necessary for microwave power devices, is addressed in this paper. High-frequency large-signal characteristics of AlGaN/GaN MODFETs measured at 8 GHz are reported for devices with gatewidths from 200 μm to 1 mm. 1-dB gain compression occurred at input power levels varying from -1 to +10 dBm as the gatewidth increased, while gain remained almost constant at -17 dB. Output power density was ~1 W/mm for all devices and maximum output power (29.9 dBm) occurred in devices with 1-mm gates, while power-added efficiency remained almost constant at ~30%. The large-signal characteristics were compared with those obtained by dc and small-signal S-parameters measurements. The results illustrate a notable scalability of AlGaN/GaN MODFET power characteristics and demonstrate their excellent potential for power applications     

Intermodulation analysis of dual-gate FET mixers

A detailed intermodulation analysis of dual-gate FET (DG-FET) mixers is presented. The analysis method is based on a large-signal/small-signal analysis using time-varying Volterra-series methods. The analysis program allows one to probe the internal nodes of DG-FETs to evaluate the nonlinear current components. Therefore, it helps physical understanding of intermodulation distortion (IMD) mechanisms in DG-FET mixers. The program was used to identify the major sources of IMD generation. It was found from the analysis that the nonlinearities due to the output conductance (G_d3 and G_d2) of the lower common-source FET were most responsible for IMD generation. The impact of the upper common-gate FET on IMD generation was also found to be nonnegligible, especially at high local oscillator (LO) power levels. The analysis also predicted the presence of MM "sweet spots" using bias optimization, which was experimentally proved by the fabricated mixers at X- and Ka-bands. The optimized X-band hybrid mixer showed measured intermodulation characteristics (OIP₃ ~13.6 dBm) comparable to those of the resistive mixers (OIP₃ ~15.3 dBm) with low LO and dc power conditions     

Straightforward and accurate nonlinear device model parameter-estimation method based on vectorial large-signal measurements

To model nonlinear device behavior at microwave frequencies, accurate large-signal models are required. However, the standard procedure to estimate model parameters is often cumbersome, as it involves several measurement systems (DC, vector network analyzer, etc.). Therefore, we propose a new nonlinear modeling technique, which reduces the complexity of the model generation tremendously and only requires full two-port vectorial large-signal measurements. This paper reports on the results obtained with this new modeling technique applied to both empirical and artificial-neural-network device models. Experimental results are given for high electron-mobility transistors and MOSFETs. We also show that realistic signal excitations can easily be included in the optimization process.     

Large-signal analysis of MOS varactors in CMOS -G/sub m/ LC VCOs

MOS varactors are used extensively as tunable elements in the tank circuits of RF voltage-controlled oscillators (VCOs) based on submicrometer CMOS technologies. MOS varactor topologies include conventional D = S = B connected, inversion-mode (I-MOS), and accumulation-mode (A-MOS) structures. When incorporated into the VCO tank circuit, the large-signal swing of the VCO output oscillation modulates the varactor capacitance in time, resulting in a VCO tuning curve that deviates from the dc tuning curve of the particular varactor structure. This paper presents a detailed analysis of this large-signal effect. Simulated results are compared to measurements for an example 2.5-GHz complementary -G/sub m/ LC VCO using I-MOS varactors implemented in 0.35-/spl mu/m CMOS technology.     

26.5–30-GHz Resistive Mixer in 90-nm VLSI SOI CMOS Technology With High Linearity for WLAN

A resistive mixer with high linearity for wireless local area networks is presented in this paper. The fully integrated circuit is fabricated with a 90-nm very large scale integration silicon-on-insulator (SOI) CMOS technology and has a very compact size of 0.38 mm$, times,$0.32 mm. Design guidelines are given to optimize the circuit performance. Analytical calculations and simulations with an SOI large-signal Berkeley simulation model show good agreement with measurements. At an RF of 27 GHz, an IF of 2.5 GHz and zero dc power consumption, a conversion loss of 9.7 dB, a single-sideband noise figure of 11.4 dB, and a high third-order intercept point at the input of 20 dBm are measured at a local-oscillator (LO) power of 10 dBm. At lower LO power of 0-dBm LO power, the loss is 10.3 dB. To the knowledge of the author, the circuit has by far the highest operation frequency reported to date for a resistive CMOS mixer. Furthermore, it provides the highest linearity for a CMOS mixer operating at such high frequencies.     

An energy-based control for an n-H-bridges multilevel active rectifier

This paper deals with the control of a multilevel n-H-bridges front-end rectifier. This topology allows n distinct dc buses to be fed by the same ac source offering a high loading flexibility suitable for traction applications as well as for industrial automation plants. However, this flexibility can lead the system to instability if the dc buses operate at different voltage levels and with unbalanced loads. Thus, linear controllers, designed on the basis of the small-signal linearization, are not effective any longer and stability can not be ensured as large-signal disturbances occur. The use of a passivity-based control (PBC) designed via energy considerations and without small-signal linearization properly fits stability problems related to this type of converter. The system has been split into n subsystems via energy considerations in order to achieve the separate control of each dc bus and its stability in case of load changes or disturbances generated by other buses. Then, a set of n passivity-based controllers (one for each subsystem) is adopted: the controllers are linked using dynamical parameters computed through energy balance equations. Hence, the system dc buses are independent and stable as experimental results demonstrate.     

Nonlinearity and cross modulation in field-effect transistors

It is shown that present large-signal models are inadequate for cross modulation in field-effect transistors, but the cross-modulation performance can be accurately predicted from a power-series approximation to the measured l.f. transfer characteristic of the device. The cross modulation for typical field- effect transistors is essentially independent of frequency up to about 100 MHz. A large-signal h.f. model of the field-effect transistor is proposed for the prediction of cross modulation and related phenomena at very high frequencies. Computer predictions of cross modulation based on this model agree well with experimental measurements.     

A distributed model of the junction transistor and its application in the prediction of the emitter-base diode characteristic, base impedance, and pulse response of the device

A distributed model for a junction transistor has been analyzed to include both dc and ac biasing effects in the active base region, with particular emphasis on a small-geometry diffused base planar transistor. For such devices with extremely narrow base width, dc biasing effects cannot be neglected. At high frequencies, the response of these devices is greatly modified by ac biasing effects which are accentuated by the significant dc biasing at large emitter current levels. Two-dimensional current flow under these biasing conditions was studied with a distributed model of the active base region. From such a model, the expressions for emitter-base diode characteristic, small-signal and large-signal base resistance, and complex base impedance valid for high frequencies have been deduced in terms of physical parameters of the devices like the geometry, base resistivity, etc. This equivalent base impedance and an ideal diode with its diffusion and emitter-base transition capacitance constitute the lumped model of the emitter-base region. For any particular frequency, the base impedance can be represented exactly by a parallel RC network. The distributed model can also predict the pulse response of the device more accurately than a lumped model and show the sensitivity of the transient response to the physical parameters mentioned above. Experimental verifications of the theoretical expressions are found to be satisfactory, and limitations of the earlier works are pointed out in regard to present devices.     

Mildly Nonquasi-Static Two-Port Device Model Extraction by Integrating Linearized Large-Signal Vector Measurements

This paper introduces a new procedure, based on linearized large-signal vector measurements, for extracting a nonlinear behavioral model for two-port active microwave devices. The technique is applied to a model structure that assumes a short-term memory condition and is formulated as a parallel connection of a limited number of frequency-weighted static nonlinearities. The proposed method consists of integrating the time-varying linear characterization of the device driven into a nonlinear state by a large signal. The experiment design and measurement setup are based on a large-signal network analyzer and are discussed in detail. In the second portion of this paper, insight is provided on the most meaningful model parameters, along with an extensive independent experimental validation, which considers a GaAs pHEMT as a case study and includes two-tone large-signal data, a wideband code division multiple access signal, bias-dependent -parameters, and dc data.     

SiGe BiCMOS technology for RF circuit applications

SiGe BiCMOS is reviewed with focus on today's production 0.18-/spl mu/m technology at f/sub T//f/sub MAX/ of 150/200 GHz and future technology where device scaling is bringing about higher f/sub T//f/sub MAX/, as well as lower power consumption, noise figure, and improved large-signal performance at higher levels of integration. High levels of radio frequency (RF) integration are enabled by the availability of a number of active and passive modules described in this paper including high voltage and high-power devices, complementary PNPs, high quality MIM capacitors, and inductors. Key RF circuit results highlighting the advantages of SiGe BiCMOS in addressing today's RF IC market are also discussed both for applications at modest frequencies (1 to 10 GHz) as well as for emerging applications at higher frequencies (20 to >100 GHz).     

SiGe pMODFETs on silicon-on-sapphire substrates with 116 GHzfmax

The dc and microwave results of Si_0.2Ge_0.8/Si_0.7Ge_0.3 pMODFETs grown on silicon-on-sapphire (SOS) substrates by ultrahigh vacuum chemical vapor deposition are reported. Devices with L_g=0.1 μm displayed high transconductance (377 mS/mm), low output conductance (25 mS/mm), and high gate-to-drain breakdown voltage (4 V). The dc current-voltage (I-V) characteristics were also nearly identical to those of control devices grown on bulk Si substrates. Microwave characterization of 0.1×50 μm² devices yielded unity current gain (f_T) and unilateral power gain (f _max) cutoff frequencies as high as 50 GHz and 116 GHz, respectively. Noise parameter characterization of 0.1×90 μm² devices revealed minimum noise figure (F_min) of 0.6 dB at 3 GHz and 2.5 dB at 20 GHz     

On the large-signal behavior of transferred-electron amplifiers with a cathode notch

Transferred-electron amplifiers (TEA) are finding applications where medium powers are required as driving stages. Most work has been done up to now under small-signal conditions. The present paper describes theoretical results of a large-signal analysis. From this analysis, it appears that different large-signal behaviors of TEA devices can be associated with their corresponding dc field distribution. We compare, by computer analysis, the evolution of the large signal impedance with the RF driving signal for different doping profiles (different stationary fields) and give design figures to increase the device added power.