A Ka-band class F MMIC amplifier design utilizing adaptable knowledge-based neural network modeling techniques

This paper describes the first implementation of an adaptable knowledge-based neural network (AKBNN) model in a high efficiency class F MMIC (monolithic microwave integrated circuit) amplifier design at Ka-band in a 0.25 /spl mu/m GaAs PHEMT technology. A single-stage amplifier based upon the AKBNN model employed shows comparable results to measured performance of a gain of 7.5 dB, a PAE of 35%, and an output power of 17 dBm.