Search

This paper presents a design methodology for a load modulated balanced amplifier (LMBA) focusing on AMPM distortion mitigation. By introducing second harmonic control as a degree of design freedom, a complex load trajectory can be selected to compensate for AMPM nonlinearities in the device without substantially affecting efficiency. A mathematical derivation is accompanied by a design procedure based on closed-form equations to fabricate an LMBA based solely on load-pull data. The theory is validated through the measured comparison of three different designs operating at 2.4 GHz in a pseudo-RF-input Doherty-like LMBA configuration, with class-J, -B, and -J* main PAs. The class-J prototype outperforms the other designs with 54% and 49% drain efficiency at peak output power and 6-dB back-off, respectively, and only 4 degrees of AM-PM across this power range. When driven with a 10-MHz, 8.6-dB PAPR LTE signal, 40.5% average efficiency is achieved with better than $-40.5$ dBc ACLR without digital predistortion.

This paper presents a study of supply modulation in a Doherty power amplifier (DPA). To validate a simplified theoretical model, a 3.5 GHz conventional symmetrical DPA using a 6-W packaged GaN pHEMT is designed for supply modulation of the main and/or auxiliary amplifiers. The DPA is characterized in CW operation over a range of supply voltages, and shows up to 15 percentage point efficiency improvement at 12 dB output back-off when both main and auxiliary amplifiers are varied simultaneously. A comparison of three cases of supply variation is shown and potential benefits for improving back-off efficiency are discussed.

We present a transfer learning method inspired by modulatory neurotransmitter mechanisms in biological brains and explore applications for neuromorphic hardware. In this method, the pre-trained weights of an artificial neural network are held constant and a new, similar task is learned by manipulating the firing sensitivity of each neuron via a supplemental bias input. We refer to this as neuromodulatory tuning (NT). We demonstrate empirically that neuromodulatory tuning produces results comparable with traditional fine-tuning (TFT) methods in the domain of image recognition in both feed-forward deep learning and spiking neural network architectures. In our tests, NT reduced the number of parameters to be trained by four orders of magnitude as compared with traditional fine-tuning methods. We further demonstrate that neuromodulatory tuning can be implemented in analog hardware as a current source with a variable supply voltage. Our analog neuron design implements the leaky integrate-and-fire model with three bi-directional binary-scaled current sources comprising the synapse. Signals approximating modulatory neurotransmitter mechanisms are applied via adjustable power domains associated with each synapse. We validate the feasibility of the circuit design using high-fidelity simulation tools and propose an efficient implementation of neuromodulatory tuning using integrated analog circuits that consume significantly less power than digital hardware (GPU/CPU).