No abstract

A nonlinear HBT table-based model is presented in which low-frequency (LF) noise effects are incorporated, thus allowing for precise oscillator design . The model uses four table-based nonlinear functions: I c , Q c , V be and Q b , and takes into account temperature and noise behaviour. This approach is able to predict the LF noise behaviour under both static and dynamic conditions by using cyclostationary noise sources. HBT-based oscillators have been designed using this model, and very good predictions have been obtained for the oscillation frequency, the output power and the phase noise.Introduction: Efficient design of microwave circuits, such as oscillators, requires the availability of accurate nonlinear device models. Nonlinear table-based models have shown accurate predictions for FETs [1] and HBTs [2], by using a simplified circuit formulation and avoiding an optimisation-based extraction procedure. However, in order for such table-based models to predict the oscillator phase noise, low-frequency (LF) noise modelling must also be incorporated. In this Letter we present, for the first time, a nonlinear HBT table-based model incorporating LF noise effects, thus allowing for precise oscillator design.

No abstract

A study is presented on the problems that may arise when characterising low frequency device behaviour with a large signal network analyser (LSNA)-based measurement system. A diode power detector has been measured and, for the first time, an X-parameters based detector model was extracted from measurements. Difficulties measuring the detector output voltage dependence with baseband impedances, especially when those impedances showed resonant effects, were observed and a method to overcome the problems encountered is presented. The measurement-based detector X-parameters model demonstrated its usefulness to predict power detector behaviour under twotone excitations and complex loads.Introduction: Diode power probes have been used successfully for many years for high-speed power measurements, when the measured signal is a single sinusoid [1,2]. In recent works [3,4] it has been demonstrated that calibrating a diode-based power probe with a singletone signal does not guarantee accurate detector calibration for complex signals. It was identified that this is related to long-term memory effects caused by the detector baseband impedances. Thus, the detector output DC voltage may vary depending on the RF input signal bandwidth.In [4], an X-parameters model for diode power probes was proposed, which will be extended in the future for calibration purposes with modulated signals. In [4], this model was extracted from simulated data. DC measurements were used to validate the detector behaviour.In this Letter, a method to perform the required measurements to extract and validate an X-parameters model for power probes using an large signal network analyser (LSNA)-based measurement system, is proposed. Difficulties encountered and possible solutions are discussed.

A load independent X-parameters-based heterojunction bipolar transistor (HBT) model has been used for the first time in the design and behaviour prediction of injection-locked oscillator circuits. This model has been extracted from load-pull measurements with a large-signal network analyser and, in order to obtain a high oscillator RF power, targeting a load impedance close to the optimum one for HBT maximum output power. A methodology is given to obtain robust injection-locked oscillator circuits with a highsynchronisation bandwidth. Several injection-locked oscillator prototypes have been designed and fabricated, and their measurements compared with the simulations obtained using the X-parameters model. Satisfactory results were obtained when the prototypes were operated as free-running and synchronised oscillators.