The noise figure (NF) of a front-end low-noise amplifier (LNA) places a lower bound on the sensitivity of a receiver. In a conventional LNA, there is a tradeoff between the intrinsic input capacitance of the input transistors and the achievable bandwidth (BW) of the amplifier. This makes it necessary to use smaller transistors at higher gate overdrive voltages to simultaneously achieve greater BW and better NF. Unfortunately, biasing the transistor in this fashion yields a power-inefficient design. Furthermore, the need for a smaller capacitance presents a challenge to electrostatic discharge (ESD) protection of the input due to its added capacitance.One possible way to overcome this noise-bandwidth trade-off is to use distributed amplifiers (DA), where the input and output parasitic capacitances of the individual amplification stages are absorbed into a transmission line (T-line) and become part of its real impedance. In a standard DA, the output currents of equally-weighted (identical-stages) transconductance (G m ) stages propagate through the output T-line and add coherently at the load. This leads to a broadband response fundamentally limited by the cutoff frequency of the synthetic T-lines. It has been shown that it is possible to design low-noise equally-weighted DAs at reasonable power levels [1].Allowing the transconductance (G m ) of different stages of a DA to be different provides us with an added degree of freedom in the design and optimization of a DA. A weighted distributed amplifier (WDA), shown in Fig. 12.3.1, can be viewed as an analog finite-impulse-response (FIR) system. Looking at the noise in such a WDA, the noise at the output is a weighted sum of noise propagating through different paths with different weights for each noise source. We can use this property to shape the profile of the output noise power spectrum by changing the G m coefficients of different stages. With proper choice of G m 's, the NF within a given frequency range can be reduced without a major penalty on the power consumption, gain, and input matching. This noise shaping can be applied to different noise sources in the circuit. For example, the noise current contribution of the input termination resistor R i to the output voltage noise is weighted by:where G m,k is the k-th block transconductance, τ is the T-line section delay, and f is the frequency. This is a more general result that reduces to the sinc function response for an equally-weighted DA. For instance, it can be used for noise profile shaping, optimization, and even dynamic noise notching at a selected frequency.A comparison of the noise contours of a conventional five-stage DA and a fivestage WDA is shown in Fig. 12.3.2, taking into account all the active and passive noise sources. In these plots, the worst-case NF across the 3-to-10.6GHz frequency band for each design is plotted vs. the gate-source voltage and total amplifier bias current for the DA and the WDA. The contours show that the WDA can provide a lower worst-case NF under the same bias conditions...