Damon G. Holmes

A simple output impedance model for Doherty peaking sub-amplifiers biased in Class C

A simple model of the Doherty peaking sub-amplifier is proposed that approximates the frequency-dependant behavior of its output impedance as seen at the reference plane of the carrier-peaking combining node. The analytical model is useful in that it shows the degradation of the output impedance over frequency as a function of fundamental device and network properties: device output capacitance, transformation ratio, and network delay. The model provides significant insight into the behavior and theoretical upper limit of output impedance when the sub-amplifier is disengaged and provides a benchmark for which to compare practical results for broadband Doherty amplifier designs. By augmenting the model with subthreshold drain-to-source resistance, excellent agreement is shown between the model and experimental results over frequency.

Behavioral modeling of Si LDMOS pre-matched devices with application to Doherty power amplifiers

Development of an experimental behavioral model to describe the non-linear characteristics of high power internally pre-matched LDMOS power transistors is presented. The model is developed by characterizing the device with swept power load-pull measurements across a large load impedance plane with the input impedance set at a pre-determined state. Based on measurements of complex input and output voltages at the devices package terminals, a functional representation is adopted to express the devices complex gain with respect to RF input voltage (or available source power) and the devices load terminating impedance. It is further shown that this approach can be used to independently model high power LDMOS pre-matched devices biased at Class-A/B and -C for use in simulating a complete Doherty power amplifier circuit. Simulations of a 300W Doherty power amplifier using this approach are contrasted to more conventional compact device models with excellent agreement noted across a large output power range. Hence, when compact models are unavailable, a behavioral one can be formulated from load pull measurements on the devices of interest.