Based on the design guidelines for low-noise HBTs which is derived from Hawkins' model, an AlGaAs/GaAs HBT with emitter-ledge passivation is fabricated for low frequency applications. This HBT increases the current gain, especially at low collector current densities--a key device parameter that impacts the noise figure at low frequencies. An NF_min of 0.6 dB with an G_a of 17.7 dB is obtained for an HBT with 2.5- X 20-micrometers X 2 emitters at 2 GHz. An AlGaAs/InGaAs HBT with regrown base contacts is also fabricated to reduce the base transit time and the base resistance, without increasing the emitter junction capacitance, for higher frequency low-noise applications. This HBT achieves an NF_min of 1.2 dB with a G_a of 10.3 dB and an NF_min of 1.7 dB with a G_a of 8.7 dB at 12 and 8 GHz, respectively, and can be useful at low frequencies as well, due to a low equivalent noise resistance that make noise matching easier.

The properties of NPN and PNP InP HBTs are reviewed and complementary HBT amplifiers built with this technology are explored. Complementary amplifiers such as those presented here manifest reduced harmonic distortion and could lead to improved wireless communication systems. The high efficiency push-pull amplifiers increase battery life, which is critical for hand-held systems. InP-based HBT technology offers distinct advantages over other semiconductor device and circuit approaches for wireless systems and the key merits of this technology are reported.

Abstract not available.

High performances have been achieved at W-band with a 2- stage, 0.1 micrometers gate-length InGaAs/InAlAs/InP LM-HEMT MMIC low noise amplifier in coplanar technology. To obtain the T- gate profile, we use silicon nitride Si_xN_y technology, which leads to naturally passivated devices. For a drain-to-source current Ids equals 350 mA/mm the devices demonstrate a maximum intrinsic transconductance Gm of 1600 mS/mm and an intrinsic current gain cutoff frequency Fc equals 220 GHz. The extrinsic current gain cut-off frequency Ft is 175 GHz. The LNA shows a minimum noise figure of 3.3 dB with an associated gain of 11.5 dB at 94 GHz.

This paper presents coplanar millimeter-wave monolithic integrated circuits with high performance and small size for use in low noise communication and radar system applications. Technology and modeling issues with respect to active and passive elements are discussed first. In a second step, the potential of coplanar waveguides to realize compact ICs is illustrated through various design examples, such as low noise amplifiers, mixers and power amplifiers. The performance of multifunctional ICs is also presented by comparing simulated and measured results for a complete 77 GHz Transceive MMIC.

This paper describes the use of an Enhancement-Depletion Mode process in Low Noise applications as well as high speed interfacing for optical networks. The same technology has also been used for highly linear, very low noise, LNAs at L- band, in down converters as well as drivers for 10 Gb/s interfaces, hence is a `Unified' process capable of performing well in many different applications. The technology will be briefly described, some of the advantages of the E-mode FET over the D-mode FET discussed and examples of circuits are presented.

This paper presents the development of a one-stage and a two-stage Ka-band monolithic LNA for commercial wireless communication applications. The monolithic microwave/millimeter-wave integrated circuits (MMICs) are fabricated with a 0.2-micrometers pseudomorphic GaAs-based HEMT technology, carried out by commercially available foundry. The one-stage amplifier demonstrated a measured small signal gain of 9.5 dB with a noise figure (NF) of 2.7 dB at 28 GHz, while the two-stage amplifier has a measured gain of 17 dB with 3.3 dB NF at 28 GHz. Due to fabrication with the commercial foundry process, these MMICs have the potential for mass production.

We report the performance of an optoelectronic oscillator (OEO) constructed with a DFB laser, a semiconductor Mach- Zehnder modulator, and a dielectric resonator based RF filter. We achieved a phase noise of -50 dBc/Hz at 10 Hz and -130 dBc/Hz at 10 kHz from a 10 GHz oscillation frequency, a performance comparable with that of an OEO constructed with a diode-pump YAG laser and a LiNbO3 modulator. The demonstration proves the feasibility of a high performance, low cost, and compact OEO.

Conventional band-pass filter design is reviewed with respect to (non-radiative dielectric) NRD-guide technology and a new step-by-step method is proposed for the design of dual-path band-pass filter using small NRD guide resonator. A mode matching formulation, accounting for inherent higher- order modal coupling effects, is used to calculate the generalized S matrix of discontinuities encountered in all kind of filters, as well as to simulate the entire filters. Upon a reduction of the transfer function of the dual-path band-pass filter, a low-pass prototype network is obtained and calculated. Based on our mode-matching software tool, an iterative optimization process is then developed to design a filter with deep attenuation poles. A practical example is given and a short comparison with the conventional band-pass filter is made.

In this paper the reliability is calculated of an all- optical C-band, 45-carrier WDM ring network comprising a large number of gain-flattened EDFA amplifiers and 10 add/drop nodes with full connectivity. A ring of the assumed type and size would cover a region surrounding the Baltic Sea, forming the top layer of a hierarchical network. Protection switching is technically uncomplicated but expensive, and may cause transmission problems due to the long distances involved. In order to estimate the reliability of the ring, the FIT number for longest optical paths is calculated and compared with requirements according to ETSI standard, a priori assuming that neither diversity, nor protection switching is available. This is obviously true for a single fiber ring in which all aggregated traffic is unidirectionally routed. From the architecture chosen follows that certain components will be crucial for the network's performance. In this context, pairs of co-working pump lasers may provide a sort of redundancy which would be a useful means to improve the reliability of the network. With this in mind, the network's resilience to non- catastrophic component failures has been simulated, with a focus on pump- and signal lasers. The global effects on optical SNR and power due to single- and multiple failures of these components are studied for arbitrary local combinations. The study enlightens a network mechanism for spectral distortion caused by pump failure, and an efficient remedy to significantly restore the signal quality.

Excellent noise and gain have been demonstrated at W-band with metamorphic HEMTs MMICs. The performance of the low noise amplifier (LNA) MHEMTs is very close to that of InP HEMTs on InP substrate. The material structure of the MHEMT has been developed using an AlGaAsSb lattice strain relief buffer layer on a GaAs substrate. DC characteristics of the 0.1 X 36 micrometers devices have shown typical extrinsic transconductance (g_m) of 1200 mS/mm to 1300 mS/mm depending on Indium mole fraction in the channel. Small- signal S-parameter measurements performed on the 0.1 X 36 micrometers devices exhibited an excellent f_T of 225 GHz and Maximum Stable Gain of 12.9 dB at 60 GHz and 10.4 dB at 110 GHz. The 3-stage W-band LNA MMIC exhibits 4.2 dB noise figure with 18 dB gain at 82 GHz and 4.8 dB noise figure with 14 dB gain at 89 GHz.

Digital application has moved towards operating speed of hundreds of Mega Hertz, with the sampling speed of ADC moving into Giga Hertz range. There is an increasing need for the design and development of a high-speed data acquisition system that is capable of capturing and processing digitized analogue signal at high speed. Due to the tight timing budget, high operating speed components, Emitter-Coupled-Logic families components with rise time of typically less than 300 ps were used in the design. With this operating speed and short rise time, signal integrity issues like reflections due to impedance mismatches and crosstalk among the traces of the printed circuit board can no longer be neglected. A quick and reliable approach was taken in the design and implementation of a 1 GHz high-speed data acquisition system using commercial-off-the-shelf discrete components. High-speed digital design issues and methodology were explored in this project and verified with the implemented hardware. This paper gives an overview of the system and focuses on the use of functional and signal- integrity computer simulation software to confirm system performance at the early design stage before actual hardware implementation. Simulation results were further confirmed with the actual hardware implemented, and was found to be close. This has helped to reduce the design cycle time and development cost of the project.

This paper presents a solution to problem of providing bit- error rate performance guarantees in a fixed millimeter-wave wireless system, such as local multi-point distribution system in line-of-sight or nearly line-of-sight applications. The basic concept is to take advantage of slow-fading behavior of fixed wireless channel by changing the transmission code rate. Rate compatible punctured convolutional codes are used to implement adaptive coding. Cochannel interference analysis is carried out for downlink direction; from base station to subscriber premises. Cochannel interference is treated as a noise-like random process with a power equal to the sum of the power from finite number of interfering base stations. Two different cellular architectures based on using single or dual polarizations are investigated. Average spectral efficiency of the proposed adaptive rate system is found to be at least 3 times larger than a fixed rate system with similar outage requirements.

A passive balun using spiral inductors and inter-digital capacitors has been developed for a 1.9 GHz balanced diode mixer on Al₂O₃ substrates. Using this passive balun on the input port (RF port and LO port) of mixer circuit and incorporating the Schottky diode, where I-V model and small signal RF model have been established, a wideband signal balanced diode mixer was designed, fabricated and measured.

The first In(formula available in paper). As micromachined V-groove gate doped-channel FET's (DCFET's) grown by GSMBE were proposed and fabricated successfully. The V-groove gate was formed by anisotropic wet etching of the undoped GaAs layer grown on top of device active layers, followed by standard metal evaporation and lift-off process. Owing to the outward slope of the sidewalls of micromachined V-groove, submicron effective gate length can be easily obtained by normal 1-micrometers UV optical contact lithography. In this way, we simultaneously achieved short gate length (for high speed) and small gate resistance (for low noise) via this `mushroom-like' V-groove gate, without resorting to the expensive and time-consuming e-beam lithography and the delicate multi-layer photoresist technique. Thanks to the inherent process simplicity, fabricated V-groove gates showed great uniformity and yield in a cost-effective way, which is essential to industrial mass production. The fabricated DCFET's exhibited a maximum current density I (formula available in paper)of 13.9 GHz at 300 K, all better than those of the counterpart DCFET's of traditional strip gates for contrast experiment. These results suggest that V-groove gate DCFET's are cost-effective devices with submicron gate FETs' performance, and are suitable for low-noise, high-speed, high-power applications.