This thesis looks into the problem of simultaneous signal and noise match at the input port of low noise amplifiers; feedback LNAs are considered because previous works show
that they can achieve the simultaneous match condition.
The investigation analyses the influence of both parallel and series feedback elements on the amplifier. Matrices are used to describe signal and noise parameters of each component of the model - parallel admittance, series impedance, active device. This approach allows the analysis to be applied to a wide range of networks, as long as noise and signal matrices are available. For this reason, the results are not limited to active devices in the microwave
region of the spectrum but they are applicable to any linear 2-port circuit.
The noise parameters of feedback networks are investigated thoroughly. Analytical expressions are worked out as functions of the feedback immittances and have been used
to support experimental evidence previously published. A duality property for feedback networks is pointed out; new circles for constant equivalent noise resistance are devised; optimum values for the feedback impedance are determined; an investigation of a well-known noise model is carried out and its validity is extended.
Based on the closed form expressions of the noise parameters, an original analytical procedure for the design of the optimum noise source reflection coefficient is presented. To the author's knowledge, no technique was available before. The design for simultaneous signal and noise match is now possible, because the input reflection coefficient can be set independently by properly choosing the load. Different devices are considered and their
different behaviour is highlighted. A remarkable feature of the new design technique is to avoid the need of input matching when designing low noise amplifiers.
Finally, experimental results are also presented and the performance of aI GHz single stage BJT LNA is shown. The fundamental achievement is that the noise figure of the LNA
is equal to its minimum value within the measurement uncertainty.