A feasibility study of broadband low–noise amplifiers with multiple amplification paths for radio astronomy

Abstract

Multipath amplifier theory: In this thesis it is proven that the theoretical minimum noise measure of a multipath amplifier (an amplifier which has multiple parallel amplifiers) is achieved by using the optimum source impedance for the amplifier and the optimum gain for each amplification path. This optimum source impedance and gain can be calculated by using the optimum–loaded input network, i.e. by replacing each amplifier with its optimum source impedance. The resulting noise measure is the same as the minimum noise measure of the amplifiers used in the amplification paths. Whereas single–path amplifiers can achieve this minimum noise measure over narrow bandwidths, multipath amplifiers are theoretically able to achieve the minimum noise measure over very broad bandwidths. The theory is demonstrated by applying it to distributed amplifiers. In an ideal distributed amplifier, the magnitude of the optimum gain of the amplification paths decreases and the phase delay increases the farther the stage is from the input, with the decrease in gain being faster for higher frequencies. The challenge in designing broadband, low–noise, distributed amplifiers is to achieve optimum gain matching over broad bandwidths. Multipath amplifier design procedure: Based on the theory, a three step design and optimisation procedure is introduced. Firstly, unconditionally stable amplification paths are designed to have small minimum noise measures, then an input network is designed for optimum source impedance matching and lastly an output network is designed for gain matching. Multipath amplifier prototype: The theory and design procedure is demonstrated by optimising a 0.5–2 GHz distributed amplifier. An average noise measure of 0.3 dB is achieved, which is only 0.1 dB higher than the minimum noise measure of the amplification stages used. This increase is mainly due to transmission line loss and gain mismatch. Radio telescope feasibility: Multipath amplifiers break the trade–off between noise temperature, bandwidth and source termination that a single–path amplifier has, because they have much more design freedom when designing the input network. In general, the more paths, the larger the low–noise bandwidth, but the larger and more complex the amplifier. Roughly two to three amplification paths are required per octave of bandwidth for LNAs around 1 GHz. When the bandwidth is very narrow, a single path is sufficient. Multipath amplifiers have similar trade–offs between linearity and power consumption, between noise temperature and noise resistance, and between noise temperature and size to a single–path amplifier. Multipath amplifiers are therefore a feasible alternative for use in radio telescopes.