Noise and dynamic range of CMOS degenerated active inductor resonators

Abstract

This paper presents a compensation method of active inductor losses, employing passive gyrator degeneration rather than an external negative impedance converter. Theoretical analysis is confirmed by a comparative simulations of two resonators designed for 434 MHz band, using Eldo RF and Spectre and the UMC 0.18 µm 1P6M process. Presented results show that the proposed loss compensation method achieves comparable noise and large signal performance to a standard active inductor resonator and provides a significant improvement in static power consumption.

Analysis and design of oscillators based on low-voltage self-oscillating active inductors

“…In the degenerated active inductor described here, three main sources can be distinguished, two from a transconductance amplifiers and the third one from a phase shifter resistor. A thorough analysis of noise properties of RC compensated resonators can be found in our previous work [4].…”

“…both phase and amplitude noise) at offset frequency is calculated. This methodology combined with the results of the detailed noise analysis from our previous work [4] and model parameters from (4), the output noise PSD of a degenerated active inductor oscillator, at frequencies ω m ≪ ω 0 is given by…”

“…This is caused mostly by a feedback in gyrator circuit where both amplifiers start to drive each other with distorted signals. Depending on the type of transconductor non-linearities, increasing signal amplitudes cause either expansion [8] or compression [4] of simulated inductance. Thus, in the presence of large signal swing, instantaneous resonant frequency of the tank varies similarly to one of a standard LC tank employing varactors.…”

“…To sustain oscillations, typically an additional negative resistance is connected to the tank. This method requires static power proportional to the amount of looses in the resonator as well as the capacitance value of the tank [4]. Thus, in the case of a low Q active inductors, standard compensation methods are not power efficient, especially with modern low-voltage CMOS processes.…”