Linearization of Response in Four-Quadrant Position Sensing Photodetector Using MLP and RBF Neural Networks

Document Type : Original Article

Authors

Abstract

In this paper, according to the size of the transistors and the feedback
gain of the oscillator, a design point in which the gain of the active
core of the oscillator is maximum, is derived. In order to realize
the feedback gain, we utilize a transformer. This transformer is
employed in such a way that it could be able to provide the feedback
gain for the oscillator and also eliminate the large portion of parasitic
capacitance, simultaneously. By using this technique, the phase noise
at 1MHz offset and the tuning range in comparison with the general
transformer oscillator is improved by 5dB and 70% respectively.
Furthermore, we show that with linearization of the oscillator in
millimeter-wave frequencies, phase noise of the oscillator can also be
reduced even more. Regarding this criteria we linearize the oscillator
as much as possible. Post-simulation results show that the circuit
designed in 0.18-μm CMOS technology, consumes 43mW while
showing -89dBc/Hz phase noise at 1MHz offset with 1.9GHz tuning
range around 57GHz.

Keywords