David B. Rutledge - US grants

This award provides funds to help with the establishment and initial operation of a microfabrication facility that will produce semi-conductor devices to be used in neuroscience, microelectronics and related areas of microdevice research. The neuroscience devices will be used for both in vitro and in vivo experimentation aimed at the simultaneous monitoring of the activity of an array of nerve cells. Other types of devices to be fabricated include micro-robots, -sensors, -antennas and pumps. An important role of the he facility is expected to be the training of students and advanced investigators who have need of these devices in the techniques used for microfabrication. The remarkable progress in the miniaturization of electronic circuit boards has lead to the realization that a variety of other types of micrometer-sized devices can be made. these include both mechanical and electronic devices. In particular, the ability to fabricate microchip-like multi-well cell culture dishes, where each well is the size of a single cell, has opened up the possibility of simultaneous recording of the activity of each of many interconnected nerve cells growing in culture. This type of experimentation could lead to significant advances in understanding the operation of networks of nerve cells, an area which has, in recent years, been the subject of much theoretical interest. Similar devices, which could be implanted directly in the brain, should allow measurement of the activity of many cells in a single region of the brain. Such experimentation has enormous potential for clarifying our understanding of the functional organization of the brain and of how brain cells integrate and differentiate sensory input.

EIA-0079871
Dimotakis, Paul E.
California Institute of Technology

MRI: Development of the Distributed Teravoxel Data System: Acquisition, Networking, Archiving, Analysis, and Visualization

This project addresses the problem of handling very large datasets through the development of generic acquisition and processing capabilities to be hosted at Caltech and made available to both Caltech and outside collaborators. Driven by investigations of flow turbulence, the project is designed to handle both laboratory and numerical-simulation data. The infrastructure will support analysis and visualization of terascale experiment or simulation datasets and their comparison to validate theories and simulation results.

0083220
Hajimiri

Most of the recent work in the area of wireless networks has been focused on systems operating at 2.4 and 5 GHz. The bandwidth available at these frequencies limits the maximum data rates for these networks. Very little has been done to use frequencies at or above 20 GHz for wireless networks. The major advantages of higher frequencies are higher available bandwidths, smaller antenna sizes and possibility of using phased array structures for directionality and gain boosting. The major impediment to operation at these frequencies is the large hardware cost, which originates from the large amount of circuitry that has to be done in gallium-arsenide or other III-V semiconductors.

An alternative approach using a large silicon integrated circuit core and a small gallium-arsenide front-end is proposed to remedy this problem. The proposed approach uses an array of resonant microstrip patches etched on a 5-cm square printed circuit board to receive and transmit signals. The antenna system can provide a gain of 20 dB, and this great directionality is available for both transmitting and receiving. The array can scan in azimuth to find, and communicate to other units on the network, while rejecting undesired interference cause by other users. The system has a gallium-arsenide monolithic microwave integrated circuit (MMIC) that only consists of a low noise amplifier (LNA) for the receive path and a power amplifier (PA) on the transmit side. Due to the high cost of gallium arsenide components, every effort is taken to minimize the size of the gallium-arsenide circuitry.

Silicon integrated circuit is at the heart of the system and is the most critical part of it. It contains the first down- and up-conversion blocks, i.e., the fully integrated frequency synthesizer operating at 21.6-GHz and two mixers down- and up-converting the signal from and to 24 GHz from a 2.45-GHz intermediate frequency (IF). This choice of IF frequency is not arbitrary and is to allow for bypassing the front-end circuitry and using a lower-frequency resonance of the array to effectively act as a single patch antenna receiving at 2.45GHz ISM band for finding other users and low bit rate data transmission in so-called sniffing mode.

The fully-integrated 21.6-GHz silicon frequency synthesizer is based on using distributed voltage controlled oscillators (DVCO), which make it possible to use transistors close or even beyond their cut-off frequency, fT. These DVCOs produce a reasonable output power with good phase noise performance and large tuning range, using a novel tuning technique, known as delay-balance, current-steering tuning. A super-harmonic injection-locked frequency dividers can be used as the prescaler to divide the output frequency to lower frequencies that can be handled by digital dynamically-loaded frequency dividers to further divide to reference oscillator frequency to maintain phase lock. To demonstrate the feasibility of this approach, a 0.35-mm CMOS DVCO operating at 10 GHz with 12% tuning range and a phase noise of -104dBc/Hz at 1-MHz offset from the carrier is designed and tested. Also two silicon bipolar DVCOs operating at 12 GHz and 17 GHz are demonstrated, further verifying the feasibility of such systems.

A complete analysis of MOS switching mixer makes it possible to exploit the higher bandwidth of the MOS transistor when used as switches and not gain devices. This approach which is based on stochastic differential equations with cyclostationary sources is used to optimize silicon up- and down-conversion mixers' performance at 24 GHz. It is shown that high performance circuitry for the first IF/sniff mode RF can be implemented in silicon, making it possible to integrate the complete back-end of the transceiver.