Engineering Clinic Projects 2020/21

Corning, Inc.

Corning, Inc. (https://www.corning.com)

Automated Gas Filter Inspection

Most exhaust filters for diesel or gasoline engines are composed of cellular ceramic honeycombs with engineered wall porosity to capture fine particles. Individual channels are open and plugged at alternating ends, having the appearance of a checkerboard on each end. Exhaust gases enter the open (inlet) channels, flow down the channel, but can only escape through the engineered porosity of the cellular walls.

Each filter contains a multitude of plugs. For example, a 13” dia. filter with 300 cells/in2 alternately plugged on each end contains >38,000 plugs. Ideally, each plug would nearly be the same length. Corning will provide the team with several representative round samples of different filter materials to aid in the development of a non-destructive method to quickly measure plug depth.

Students can use whatever techniques they desire (mechanical, electrical, acoustic, laser, etc.) to determine plug depth for each plug on each face of the ceramic filters. The desire would be to map the entire face of the filter and ideally provide an absolute length for each plug (± 1 mm) or, at a minimum, identify plugs that fall outside of some parameters – plugs are longer than X, but shorter than Y). The team could evaluate several alternative methods, against appropriate criteria, then successfully demonstrate their chosen method.

Team: 5 ECE or 3 ECE and 2 CS/CE

L-3 Communications

L3 (https://www.l3t.com)

Named Networking

Mobile networks used in disaster recovery scenarios have some unique challenges.  Until infrastructure like cellular networks can be restored, establishing communications can be difficult.  The problem is compounded by the number and variety of organizations that show up to help.  Each may bring with them their own mobile wireless system, but connecting them together is complicated by the fact that there can be no IP address duplication between them and (just like home computer networks) the probability that they have assigned their addresses from the private IP address space is very close to unity.  To allow simple interconnection, we need to solve the problem of data forwarding between (and perhaps within) mobile networks in spite of IP address duplication.  Some work has been done in this area under what is called Content Based Networking, whose paradigm is not to push data towards a destination, but rather for nodes to (anonymously) subscribe to published data flows.  The objective of this senior project is to put together a testbed for exploring solutions.  The initial target is to port the algorithms of Named Data Networking onto either a collection of Raspberry Pi or a simulation environment such as OPNET or the EMULAB environment at the U of Utah.  This will allow future efforts to explore the issues of scaling and managing a publish-&-subscribe forwarding paradigm in concert with the destination address based forwarding paradigm of IP networking.

Team:  4-5 CE/EEs

Advisor:  Mingyue Ji (https://faculty.utah.edu/u6007330-MINGYUE_JI/hm/index.hml)

Raytheon - Machine Learning for MIMO Multipath Mitigation

Machine Learning for MIMO Multipath Mitigation

An example of multi path occurs when line of sight (LOS) and bounced copies of a signal combine 180° out of phase, destroying the signal. MIMO (Multiple In Multiple Out) phased array antennas offer new opportunities to fight multi path. Different techniques investigating phased array antennas will be investigated. Machine learning can combine known, and discover new, methods to mitigate multi path. Both methods will be investigated for this project.

Deliverables include:

  • 1. Derivation of equations relating antenna geometry, transmitter and bounce locations, frequency, and phase to multipath destructive interferences for a wide bandwidth signal.
  • 2. Code creation for a simulator for multipath signals as received at each element of the MIMO array.
  • 3. Creation of a real multipath signal using the POWDER network. Calibrate array with the narrowband signal.
  • 4. Develop a metric or metrics for best multipath mitigation.
  • 5. Training a neural network to mitigate multipath with a small number of elements.
  • 6. Extra Credit: Use the elements in the MIMO array and/or mitigate two signals at once.

Note: your final presentation will be presented at the technical open house AND at local Raytheon office.

Team: 4-5 CE/CS/EE

Raytheon - Monitoring and Measuring of POWDER Network

POWDER (Platform for open wireless data-driven experimental research)[1] is a $100 million state of the art platform at the University of Utah use for development and testing of the next generation 5G wireless networks. While POWDER development continues, the network, as is today, already offers a wide range of features and functionality. Raytheon Applied Signal Technology (RAST)[2] Provides a wide range of solution for spectrum intelligence, surveillance, and reconnaissance for customers. Our local office leverage is the talents of 100+ engineers to build and maintain many products, including a product that is used to design network systems and monitor their performance.

The objective of this project is to use AST tools to model POWDER components and then monitor and characterized the performance of POWDER. Doing so will provide capabilities useful for understanding nuances within the network and to characterize performance through the visualizations and through data. The ultimate goal is to understand how to maximize overall performance of POWDER. Milestones in this project would include:

  • 1. Use C2I Designer (an AST network design and modeling tool) to build a model of the POWDER network. This would require the students to understand the hardware topology on POWDER and then to add hardware support to the AST tool where hardware pieces are not currently supported.
  • 2. Use the C2I and Gemini clients (AST tool and laboratory monitoring and visualization tools) to import the model and then capture and view spectra from the POWDER network. Virtual sensors will need to be created to send the data to C2I.
  • 3. Extra credit: use machine learning to find POWDER lab operational outliers in the state of health data that is being collected by the tools.
  • 4. Extra credit: use machine learning and\or statistical techniques to discover the appearance and disappearance of RF signals in the spectra collected by the tools.

Note: your final presentation will be presented at the technical open house AND at local Raytheon office.

[1] https://powderwireless.net
[2] https://www.raytheon.com/capabilities/products/ast

Desired Student Background:

  • C++(preferred), Java, or object oriented software development
  • JavaScript and Angular software development, or similar
  • Computer Networking (especially wireless, and 5G knowledge even better)

Team: 4-5 CE/CS/EE

Rocky Mountain Power

Distributed Battery Storage Control
Customer owned distributed energy storage is becoming a ‘new normal’ and presents an opportunity for utilities to integrate customer and utility storage into their power dispatch procedures. Projects such as the Soleil Lofts and the Panguitch battery energy storage system reveal that utilities will have to accommodate multiple vendor software control systems to effectively dispatch, leverage, and coordinate customer and utility energy storage. The proposed project is to develop an energy management application that is capable of monitoring and controlling various energy storage systems over DNP3 communication. The application will operate in a simulated environment and will need to accommodate trigger events to respond to situations such as load shifting, voltage regulation, and power transfer market requests. The project will provide PacifiCorp a framework of how to streamline energy storage management to a single user interface in lieu of dedicated systems to each storage system.
Team:  3-5 EE/CE, 2 CS
Advisor: Angela Rasmussen

Sandia National Laboratories

Debugging a faulty RISC-V core

What happens to secure software when the hardware it’s running on experiences a fault? Can it be relied upon to safeguard your secrets, or will it be coerced into spilling them? What novel techniques can we develop and test to mitigate against this threat? Project Velocity builds on a previously-developed accelerated fault testing platform to help answer these questions. Based on the RISC-V processor architecture, we plan to implement and test various hardware and software mitigations against hardware faults, with the goal to develop low-overhead security countermeasures.

Team:  4-5 CE
Advisors: PE Gaillardon and Jon Davies

Electrical and Computer Engineering Faculty Projects

Mike Scarpulla

Optoelectronics/Solar Cells/Semiconductor
Processing

Sub-Project 1:

Design of methods to simulate optoelectronic device characterizations – The performance of many optoelectronic devices such as solar cells, LEDs, lasers, as well as traditional CMOS devices depends critically on the behavior of dopants and defects in the semiconductors and at interfaces such as contacts. In this project, we will explore the use of device simulation programs that solve the drift-diffusion and Poisson equations to simulate the transient behavior of CdTe solar cells and GaN diodes. In both cases, my research group has observed time-dependent hysteretic behavior possibly due to charge trapping. Thus, these simulations will be compared to actual ongoing research data to try to determine the root causes of the observed behaviors.

Sub-Project 2:

Design and implementation of automated test and measurement and data processing workflow – In one project, we are measuring I-V curves on dozens of individual GaN diode devices in-situ during nuclear irradiation. Similarly, we do testing before and after on up to 72 diodes and multiple TLM test structures on each sample measuring only a few square cm. We have automatable hardware that can switch between the devices (which will be wire-bonded to PCB boards) and execute sequences of tests such as I-V, CV, etc. After the data is measured, we need to process it and produce results – for example extracting the turn-on voltage for diodes or the bulk and contact resistances from the TLM patterns.

Sub-Project 3:

Design and implementation of bias and amplification circuits for Ge photon detectors  – Biased ultrapure Germanium detectors are amongst the best ever for radiation and light. We require a custom designed bias and amplification circuit to be designed and built to bias the Ge detector itself and amplify the signal coming from a JFET-based preamplifier. Circuit designs are available, the original board can be reverse-engineered, and some preliminary work has been done by a prior student.

Sub-Project 4:

Spectrometer Design. One of the following possible projects;

  • Design and building of a spectrometer based on fixed diffraction grating and InGaAs CCD camera –  fiber coupled spectrometers based on Si CCD arrays have supplanted spectrometers based on monochromators (an optical system with rotating diffraction grating but one detector). However, in the near IR InGaAs detectors are far superior and extend the sensitivity range to 1500 nm. An existing InGaAs CCD camera should be integrated into an optical system to allow analysis of light from optical fibers in the 900-1500 nm range to be analyzed and turned into spectra (counts vs wavelength, with each pixel representing a range of wavelengths).
  • Designing data processing routines for semiconductor defect spectroscopies – In this project, students will work in Matlab to design methods of processing data from DLTS and related experiments. The raw data is typically a decaying exponential of capacitance vs time. Data is taken vs temperature, resulting in a range of decay times from fast to slow. A set of correlation functions with different time signatures is used to determine a set of output signals vs T which yield signal vs T peaks. Then, these center of these peaks is determined and Arrhenius analyzed to determine activation energy and capture cross section.

Sub-Project 5:
Semiconductor Processing. One of the following possible projects:

  • Designing data processing and analysis routines for transient photocapacitance data – Transient photocapacitance involves exposing a semiconductor sample to fast step changes of light (i.e. opening or closing a shutter) and then measuring the change in capacitance vs time. Similar to the DLTS project above, these transients must be analyzed in terms of their decay time constants and then that data analyzed vs wavelength and temperature.
  • Design of epitaxial growth processes for Cd3As2 on III-V wafers – this project will involve designing procedures to use a Molecular Beam Epitaxy (MBE) system to grow single crystal epitaxial layers of Cd3As2 heteroepitaxially on GaAs, GaSb, and/or CdTe. Parameters such as crystallographic growth axis, surface preparation, temperature, growth rate, and stoichiometry of the fluxes must be optimized.
  • Design of optimal contacting processes for Ga2O3 – Ga2O3 is a material of interest for high voltage electronic devices. It can be doped fairly readily n-type, but producing ultra-low resistivity Ohmic contacts as well as Schottky contacts requires optimization of the metal deposition parameters and post-annealing.  Surface cleaning procedures will be tested. Deposition by sputtering and e-beam evaporation will be attempted. Annealing in oxidizing and reducing atmospheres These integrated processes will be designed and tested by this team.
  • Designing a remote plasma system for activating O2, H2, and other species during annealing – The chemical activity of gases is typically conceived to consist only of its partial pressure and temperature. However, an additional chemical activity can be imparted by exciting a gas atom or molecule by addition of light or other EM radiation. Plasmas are frequently used to crack molecules and produce excited species with higher chemical activity. In this project, participants will design, build, and test a plasma source for use in tube furnaces. This will consist of either a capacitive or inductive coupling scheme (i.e. plate-like electrodes or a coil, both being outside of the quartz tube).
  • Design of illuminated annealing setup for Ga2O3 – Students working on this project will design an illuminated annealing process to modify point defect populations in Ga2O3.  The material will be annealed at temperatures >900 ◦C and illuminated with UV photons. This will be tried inside an existing vacuum chamber and/or inside furnaces.

Sub-Project 6 (outreach):
Design of ultrasound diffraction demonstration system – Students will design and build an ultrasound diffraction system for use with crystals and polycrystals. The diffraction system will be used for outreach and educational purposes. It will consist of periodic arrays of rods held under water, upon which an ultrasound beam will impinge from a source. The diffracted waves will be measured with an ultrasound detector. Both the source and detector will be movable in angle, analogous to the source and detector of an XRD system.

Mingyue Ji

Federated Machine Learning for Computing

This project is based on IBM research. This project delves into the field of Machine Learning (ML) applied to low computing power devices to perform data recognition and prediction, as compared to the wide spread, virtually limitless, computational capacity of cloud computing.

Heayoung Yoon

Three-Dimensional Microstructures for Emerging Optoelectronic Devices

Electrode arrays are critical for interfacing with the human nervous system, and despite years of use in medical and research applications, the understanding of the characteristics of implanted electrodes is incomplete. It is necessary to design better measurement tools to improve our understanding of the electrode-tissue interface. Students in this project will be expected to design a data acquisition system including the analog front end, digitization of the analog signal, USB communication, and MATLAB or Python software to save the data and control the data acquisition system. We are interested in two students willing to start in the summer and we will help them apply to UROP to secure funding for Fall and Spring semester.

Team: (1 person)

Armin Tajalli and Berardi Sensale Rodriguez

Circuits and Devices for Optical Communications

The goal of this project is to design high-performance and high-speed circuits and devices for modern optical communication systems. Such communication links are supposed to transfer data between different processor units in a high-performance computing system, and thus improve their processing power. Serial data communication is a very common way of moving data within processors, or between different cores of a processor and memories. Processing power of the new computers depends highly on how fast the data can be moved between different units of a multi-core system. Because of that, communication data rate has been continuously increased. The next generation serial data communication lanes are supposed to carry more than 112 Gb/s/lane. As a conclusion, transferring data over fiber optics has became very attractive. Students involved in this project, will learn about fundamental of modern optical communications, optical devices and structures (such as optical detectors, modulators, waveguides, among others), design of circuits that can operate with optical devices, optical system architectures, and many more theoretical and practical topics related to these type of systems. This project will be a very valuable experience for the future carrier of students, specially those who want to continue in the field of communications, optical systems, and integrated circuit design.

Background: Students with a good background on circuit and device design are encourage to apply. Also, students need to be self motivated and active. Good knowledge on relevant Software Tools (e.g. Matlab, Python, SPICE) is a plus.

Team: 2 to 4 EE/CEs

Advisors: Prof. Armin Tajalli (armin.tajalli@utah.edu , www.lcas.ece.utah.edu ) and Prof. Berardi Sensale Rodriguez (berardi.sensale@utah.edu)

Armin Tajalli

High-Speed Analog to Digital Converter Circuit

The goal of this project is to develop a methodology to optimize design for very high-speed analog-to-digital converter (ADC) circuits used in RF circuits. Data converters are key building blocks to impairment communication systems. Modern RF 5G systems are using data converters to convert signal to digital for further signal processing. Also, data converters are used in serial communications in high-performance computing systems.

In this project we are planning to develop a design methodology in order to implement very high-speed analog amplifiers and ADC circuits, estimate their performance and power dissipation, and propose an optimal design. Based on that, a software tool will be developed, which can offer an optimal design depending on technology of IC fabrication, speed of operation, or target application. We will use the tool to make a practical design and implementation the integrated circuit as well. Students involved in this project, will learn about fundamental of deign of high-speed amplifiers, data converter systems, advanced techniques to design, model and analyze such systems, and learn about analog IC layout techniques. Thus, such a project will certainly be a very valuable experience for students for their future carrier. Interested students can move forward and participate in design of such systems and work on cutting edge projects.

Background: Students with a good background on circuit design are encourage to apply. Also, students need to be self motivated and active. Having good knowledge on relevant tools (e.g. SPICE, Matlab, Python) is a plus.

Team: 2 to 4 EE/CEs

Advisor: Prof. Armin Tajalli (armin.tajalli@utah.edu , www.lcas.ece.utah.edu )

Rajesh Menon

Deep-learning to enable imaging for autonomous agents

By applying machine-learning techniques to image sensor data, autonomous agents can extract more information than might be accessible to humans and thereby, make more accurate and informed decisions. In this senior project, the student team will: (1) create software based on machine learning algorithms; (2) perform simple optical experiments (guided by graduate students in most cases) and (3) characterize the efficacy of the machine-learning algorithms. The students must have some experience with machine learning and a very strong willingness to learn and work independently. Successful outcome includes publication of peer-reviewed journal article(s) and prototype demonstrations that will be seen by major recruiters.

Background: Need to have previous experience with machine learning.

Team: 1-2 CE/EEs

Advisor: Prof. Rajesh Menon

Florian Solzbacher

Renewable Energy

A novel manufacturing methodology will be investigated for renewable energy. This will involve research for thermomotive linear actuators. As the earth changes its temperature every day (diurnal temperature variation), we can use that temperature change to generate mechanical work and convert it into electricity.

The following is a list of milestones and/or deliverables for this project:

  • 1. Manufacturing of optimal Twisted Coiled Polymer Actuators (largely predetermined with textile engineering)
  • 2. Housing for storing the actuators
  • 3. Converting mechanical work to electrical power
  • 4. Passive regulation of environment
  • 5. microgrid regulation in collaboration with Dr. Masood Parvania

Team: 2-4 CE/EEs

Register for projects at the registration page.