Low-Noise Components

Rugged Low-Noise Amplifiers

The signals received through a wireless link are commonly very weak. Therefore, low-noise amplifiers (LNA) are used in the first receiver stage in order to amplify the signal to a power level required for further processing without adding significant distortion.

Although LNAs are designed to amplify very low power levels, there are chances that they receive very strong signals, which is not desired. This could happen for example if receive and transmit units are placed close to each other, like in mobile base stations. The traditional approach to prevent damage of the LNA due to input overdrive is to place power-limiting circuits between the antenna and the LNA.

LNAs based on GaN transistors, on the other hand, have the potential to withstand high input overdrive powers without requiring protection circuitry, since the material system intrinsically provides high breakdown voltages and high-power handling capability.

The FBH is currently developing highly rugged LNAs for space applications. A LNA for 3,5 - 7 GHz providing  a noise figure below 1.8 dB has been demonstrated. It survived an input overdrive power of 33 dBm (2 W) for 16 h without any degradation [Publication, Rudolph 2006], [Publication, Rudolph 2007]. Commercially available LNAs based on GaAs technology, for comparison, are commonly specified for maximum input powers below 20 dBm (100 mW).

Low phase-noise Oscillators (VCO) up to 80 GHz were realized in the FBH GaAs HBT MMIC.

The oscillators feature record values among mm-wave oscillators concerning their phase noise.

Accurate transistor models are required in order to be able to simulate electronic circuits. These models need not only to describe the general electrical performance, but also the noise properties of the device. Concerning heterojunction bipolar transistors (HBTs), the common models were not satisfactory. The FBH HBT large-signal model provides a number of features that improve the state-of-the-art in HBT noise modeling and which are based on the following research results:

• low-frequency (1/f) noise
  • Two 1/f noise sources are required and sufficient for an accurate model. The noise sources can reliably be determined from measurement [Publication, Heymann 2001].
  • The 1/f noise is not only determined by the DC current, but also by large-signal RF currents. The 1/f noise sources therefore become cyclostationary noise sources in the large-signal case [Publication, Rudolph 2004], [Publication, Rudolph 2006].
• Shot noise
  • The correlation of the shot-noise sources must be considered [Publication, Rudolph 1999].
  • The correlation time constant can be approximated by the collector transit time. The noise correlation can be approximated by proper placement of two noncorrelated white noise sources within the equivalent circuit. A dedicated nonlinear shot-noise model is thus not required [Publication, Rudolph 2007], [Publication, Rudolph 2007], [Publication, Rudolph 2008].