Project

60-GHz CMOS Radio Systems

Introduction

The aim of the OGRE group is to design a fully-integrated low-power multiple  antenna wireless LAN radio front-end at 60 GHz using conventional silicon technology.

mm-Wave Devices are “exotic” and therefore   expensive ......... How far can we push CMOS and Bipolar?

This will enable low-cost, low-power, mobile, agile mm-wave communications.

A broadband wireless system capable of providing 10x capacity and operating at 10x the carrier frequency of current wireless radios will require a fundamental shift in the design of CMOS circuits and new approaches to system design.  To address these challenges, our research is focused on several key areas:

mm-Wave  Circuit Design

	Chip photo of a 28 GHz SiGe low noise amplifier

Designing circuits operating at mm-wave frequencies using a digital CMOS process poses many challenges due to the lossy substrate, low f_t and f_max, and noisier devices. As the wavelength approaches the size of on-chip dimensions, new design methods incorporating microwave techniques and complex passive structures can be used to improve circuit performance. In particular, techniques using transmission lines, distributed elements, and matching networks are being investigated and applied to the design of typical transceiver building blocks such as the LNA, VCO/PLL, mixer, and PA. The ultimate goal is to realize a 60GHz CMOS transceiver that directly integrates with the baseband circuitry.

A separate transceiver employing SiGe HBTs is being developed in parallel in order to implement the front-end more quickly. 

Device Modeling

	Chip photo of a passive device test structure

The design of circuits operating near the frequency limits of a given technology will require accurate active and passive models. Presently, none of the available CMOS transistor models are verified at mm-wave frequencies. Furthermore, at 60 GHz, layout-dependent parasitics, which could be neglected at low-GHz frequencies, must be taken into account. Equally important is the characterization of passives on lossy silicon substrates--transmission lines, inductors, capacitors--which are used for resonators, matching networks, and bias circuits. We are continuing Berkeley's long tradition of CMOS modeling, and extending it to mm-wave frequencies!

Microwave Measurements

	BWRC is equipped with an advanced  on- chip mm-wave measurement setup.

Accurate on-wafer mm-wave measurement is an integral part of our modeling and design flow. We use automated network analyzers, high-speed wafer probes and programmable DC source/monitor units in conjunction with state-of-the-art calibration techniques to characterize mm-wave active and passive devices on lossy silicon substrate.  Employing advanced characterization and measurement control softwares such as ICCAP enhances speed and accuracy of our measurement-modeling flow. 

Owing to a special extension to BWRC’s RF noise measurement set up and standard VNA hardware, we will be capable of measuring performance of fabricated circuit blocks such as LNA, mixer, oscillator and power-amp up to 65 GHz at BWRC.

Radio Architecture Design 

Conventional  transceivers consist of three main parts: RF front-end, analog core, and  digital signal processing unit, with a clear separation between lumped circuits on-chip and limited consideration of distributed effects off-chip (package and board). They also comprise many off-chip front-end components (filters, switches, networks, antenna). On the other hand, our envisioned 60GHz transceiver system takes advantage of the possibility of integrating passive components as well as arrays of antennas directly in the package. Since many of its structures are electrically large, they are designed as distributed circuits.

The need for relatively small size antenna and passive structures at 60GHz allows the direct integration of antenna arrays as well as complex passive structures with the active circuitry onto the same substrate. The implementation of antenna array topologies results in increased channel capacity as well as resilience to multi-path fading due to spatial diversity and improved antenna gain due to beam forming.

Wireless System Design

The lossy and multipath nature of the 60GHz wireless channel, coupled with the limitations of conventional CMOS technologies at this frequency, makes high-SNR communication in the 60 GHz band extremely difficult. Free-space path loss is increased ~10x relative to 5 GHz communications, and 60GHz silicon components have limited noise and power-handling capabilities. Multi-antenna approaches (described above) can directly combat some of these issues in order to achieve a better channel characteristic and better SNR. The 60 GHz band, however, has 7 GHz of usable BW, so a high-throughput system does not necessarily require high spectral efficiency (and hence, high SNR).

We are investigating system algorithms and architectures that are both amenable to the limited capabilities of conventional silicon technologies and capable of overcoming the challenges of the 60 GHz channel.  The low required spectral-efficiency required for 60GHz operation allows for investigation into alternative system architectures that are better suited to conventional silicon implementations. Also, the significant processing capabilities of digital CMOS architectures allow extensive estimation of and compensation for the inherent impairments of silicon technologies in the 60GHz band.