Summer
Research Experience for Undergraduates (REU)
Michael Ashfield - Micro-Ellipsometry, Instrument Control, and Signal Detection
S. Riki Banerjee - Realization of a 1675 MHz Low Noise Amplifier Design
Ryan Cobian - Fabrication and Characterization of Integrated YIG Waveguides
Marcia Golub and Matthew Legler - Nanoelectronic Logic Circuits
Zachary Handahl - Affects of Flow Velocity on the Diffusion Rate of Oxygen in the Blood
Kelly McQuerry - Mapping Visual Fields
Nancy Meares - Superthreaded Computer Processor Design
Brad Pokorny - Designing a Blood Pump Assist Device
Tyrone Roach - Performance Analysis of Cellular Radio Systems Over Fading Channels
Nicholas Sherwood - Tunable-Coupler Based Semiconductor LASER Design
Michael Sutton - Performance Evaluation Using an Eigenvector Learning Method for Mobile Robots
Serena Thompson - Digital FM Signal Recognition for ETAH (Electronic Totally Artificial Heart)
Brian Usner - An Optimal Variable Structure-Sliding Mode Design for a SISO Nonlinear System
- 2000 Summer Program Staff
Professor Douglas Ernie Professor Lori Lucke
2000 Summer Program Abstracts
Technology today is focused on the fabrication of smaller and thus faster devices. Ellipsometry is an optical technique that is important in metrology and can be used for small feature analysis. Its non-destructive and highly sensitive features make ellipsometry ideal for use with any materials. Conventional ellipsometers have drawbacks in that they are unable to resolve very small features of a sample. Using a high numerical aperture (NA) objective lens the ellipsometer can now create a highly resolved image of the sample. This resulting design is a radially symmetric ellipsometer. In this project, LabVIEW was used to implement a user interface to control the apparatus and handle data collection.
A low noise amplifier was designed for 1675 MHz using a commercially available MESFET product (Agilent MESFET ATF-10236), microstrip lines, and appropriate chip components. Previous design work including calculations, a parts list, board layout, and initial simulations was available from another student. The goal of this project was to realize this design, including validation of the design, prototyping, and testing the LNA performance using a network analyzer.
Magneto-optical isolators and switches need to be developed on semiconductor platforms for use in Photonic Integrated Circuits (PICs). These are circuits that use light instead of electricity. In order for one-chip isolators and switches to be a reality, magneto-optical materials such as Yttrium Iron Garnet (YIG) need to be integrated onto semiconductor substrates. This project focused on the deposition of YIG onto magnesium oxide (MgO) substrates and on the characterization of the magneto-optical properties of the YIG waveguides. MgO has been shown to be a good buffer layer for semiconductors. Deposition of YIG onto MgO substrates was attempted via reactive rf sputtering. A yttrium target patterned with pieces of iron foil was used to deposit the samples. However, the target did not have enough iron, so the samples were iron deficient. The samples were characterized with X-ray Diffraction, Energy Dispersive X-ray, and a Vibrating Sample Magnetometer. An experimental apparatus was also set up to measure Faraday rotation in thin film waveguides. Faraday rotation is the rotation of the polarization of a light wave inside a magnetized magneto-optical material. This apparatus was capable of coupling laser light in and out of thin film waveguides and was sensitive enough to measure 0.05 degree rotations.
This research centered on the development of new circuit technologies in the area of nanometer scale devices. The fundamental limits of current circuit technology will be reached in the not so distant future if circuit miniaturization trends continue at their present rate. As such, new ways of building circuits need to be invented if we desire the trend in faster, smaller, more powerful computing to continue. This project was concerned with the testing of a new form of computing logic required for one such possible avenue in this advancing area; primarily tunneling phase logic (TPL), which is a reemerging way in which logic can be performed. It is a way to represent logic using the locked phase of single electron tunneling events. This deviates from traditional devices since the phase of a waveform is used to represent logic levels instead of high and low voltage values. In testing this form of logic, a single electron circuit simulation package called SIMON was utilized. Testing of TPL has been performed in the past and is still being conducted. This project improved upon the database by more accurately simulating the conditions involved with signal transfer within nanoelectronic devices. More accurate simulation was achieved by using a sawtooth input waveform instead of a sinusoidal, since it closely mimics the output waveform peak voltage values. Upon testing of the sawtooth waveform, we found that it functioned in much the same way as the sinusoidal waveform in respect to a nanoelectronic array.
We have made definite progress towards fabricating a very inexpensive planar waveguide optical amplifier that can perform comparable to an Erbium Doped Fiber Amplifier (EDFA). Improving upon previous ridge waveguide designs, a trench waveguide design was developed that should maintain gain and decrease loss. We encountered the problem of trench shadowing due to the directional nature of ion beam deposition; this problem was solved using a Reactive Ion Etch (RIE). We developed appropriate RIE recipes that yielded the desired geometries for Alumina and Zirconia waveguides. Subsequent RIE etching roughness problems were overcome using an oxide polishing etch.
Researchers at the metabolic surgery research laboratory at the University of Minnesota are currently studying the effects of cholesterol on the rate of the diffusion of oxygen in blood. It is believed that cholesterol causes the membrane of red blood cells to become rigid and therefore decrease their ability to pass oxygen. The researchers at the lab are testing this theory and needed assistance in a certain aspect of the project. More specifically, they wanted someone to research the effects of blood flow velocity on the diffusion rate of oxygen in the blood. I was happy to accept this task as my summer research project. At the conclusion of my project, the research team can then use my results to aid them in their study of cholesterol.
Fibre Channel is an integral part of a recent revolution in data communications. The new developments is graphics technology, interactive computing, long-distance file sharing and other large scale, large bandwidth applications often require very high speed interface links to serial communications and storage I/O. Fibre Channel architecture can provide solutions to all these problems and more. However, much research and experimental work remains to be done in order to achieve all the goals set for Fibre Channel.
Current methods for mapping visual fields only determine if a patient can see in a particular location, not what the patient sees there. By advancing the tests used to map visual fields, one can start to understand more about what we see, how we see it and why we see it. Our visual fields tests will determine where in the visual field one can accurately differentiate orientations of lines. We will also investigate how and where in the visual field color and shading are seen.
The superthreaded processor is a computer processor that uses multiple lightweight processor on the same chip. The processor is in pre-fabrication stages and is being tested using both a hardware-based simulator, and software based simulator. The software based simulator is called SIMCA. SIMCA is being tested with SPEC benchmarks. However, the SPEC benchmarks are written to be tested with the fabricated processor, and would take years to test under SIMCA or most other simulators designed for simulating future architectures. Thus, the benchmark datasets need to be reduced in a defendable way. I have been working with one of the CPU 2000 benchmarks and one of the CFP 2000 benchmarks in reducing the datasets in such a way that a profile of the reduced set reflects the original accurately.
A Blood Pump Assist Device helps the heart rejuvenate its strength when it has become weak. The device is attached to the body in the left leg and near the left shoulder and pushes blood through the body using a centrifugal pump. This is especially useful when treating ailments such as heart disease, in which a limited amount of rest can improve the heart's health a great deal. We have had to keep in mind certain conditions that must be met in addition to making a functional device. We need to create this product in a way that is timely, cost effective, and satisfies the wants of our customer. While designing Fore Flow's user Interface, communication with our client and keeping the interface intuitive have been two important aspects that have influenced the design process. Another major part of the design process, choosing the processor for the device, has required an analysis of Fore Flow's processing needs so that we ensure the processor will be able to support all of Fore Flow's components.
Unlike wired communication, wireless communication is not predictable. The advent of the cellular concept helped increase spectral efficiency and user capacity in regards to the mobile telephone. However, random signals generated by a mobile phone system are subject to many hostilities of the environment, e.g., fading, multipath, Doppler shift, and co-channel interface; many of which can seriously degrade its performance. Moreover, these signals tend to behave in distinct ways such that their histograms are mapped to probability models. This area of research is concerned with Nakagami and Ricean models and thus focuses on quantifying the performance of cellular mobile systems efficiently over the above mentioned fading channels. Programming tools such as Mathematica are used to compare closed and numerical form derivations for computational efficiency and accuracy. Additionally, a graphical analysis is done to test the mapping between Nakagami and Ricean models for arbitrary parameters. It turns out that the closed form is ideal to have. It is more efficient and simpler than the numerical form. Furthermore, differences exist in the distribution mapping; however their significance is dependent upon the application.
This research project has the objective of creating a metal cutting tool that utilizes a small semiconductor laser as its power source. The advantage of this is that unlike present metal cutting devices it will not require a bulky high-power laser to produce a good cut. The coupling action of our devices acts to prevent reflection from metal or other highly reflective surfaces such that 100% light absorption can be achieved. A multilayer distributed Bragg reflector made up of various dialects will be mounted on a silicon or glass substrate and will utilize a thermal bimorph actuator, which will produce vertical displacement of the mirror by means of heating the element using a small current. The whole device will be placed on a small microscope objective, which can be focused manually on any sample to be evaporated. During this summer's project, one full set of devices was designed, optimized, built, and tested. Some problems arose in the manufacturing process, but results were overall positive. A second design, improving on the first one's flaws in processing was designed, but remains untested.
We present a method to evaluate the performance error of an eigenvector learned sensor/actuator mapping for mobile robots. During runtime, the learning system projects sensor data onto the eigenspace using eigenvectors determined in training. The result of the projection is a set of sensor values and actuator values. We develop an error metric based on comparing the projected values with the actual sensor values. When the system performs closely to how it was trained, the differences between projected and actual sensors are small and hence the error metric is small. The error increases as the system fails to perform how it was trained. Unlike many other self-evaluation techniques, this method is not task specific and can be used for any eigenvector learned skill. Two examples applications of the error metric are shown using wall following skills for a mobile robot. First the metric is used to determine transition points in sequential tasks. Then the error metric is used to create an adaptable robot that chooses the best performing skill given the sensor readings.
Replacement of the human heart by artificial hearts has been studied for decades. One of the designs closest to completion is the electric totally artificial heart (ETAH). A key component of this device is an electronic controller, which can be monitored and programmed remotely through a telemetry link with a PC. This project investigated possible methods for improving the performance and reducing the cost of the telemetry link. The methods studied employed a programmable logic device (PLD), using simple logic, to recognize valid data received by an inexpensive commercially available digital radio kit. This technique alleviated detection problems due to static noise inherent in the radio's design. Three alternative PLD designs were simulated to verify performance. The two best designs were prototyped and their operational suitability confirmed.
When an engineer builds a mathematical model for a given system, there tends to be mismatches due to unmodelled dynamics, variation in system parameters, or the approximation of complex system behavior by a straightforward model. The engineer must design a controller that is robust to all of these discrepancies. One approach to robust controller design is the sliding mode control methodology. My project presents the design of a variable structure-sliding mode controller for a robotic arm. Due to nonlinearities, feedback linearization techniques were used in order to construct the sliding mode controller. The sliding surface of the controller was designed to minimize a quadratic cost criterion. Because of chattering effects, the signum function was replaced by a typical smoothing function in order to achieve the desired results.
