Undergraduate Capstone Project

2026-04-11T00:00:00+00:00

Undergraduate Engineering Capstone</h1>
Engineering capstone is the final undergraduate engineering project students complete in their undergraduate degree. Teams of 6 to 7 students are assigned a project provided by local industry partners or university faculty and completed over the span of their final year (2 terms).</p>
The project assigned to my team was titled "Development of miniaturized readout electronics for microwave ice sensing applications in harsh environments".</p>

Project Overview (The Why</em>)</h2>
Ice buildup is a serious issue in aerospace and wind energy systems, where accumulation can reduce aerodynamic performance and create safety hazards. Our capstone project focused on designing a portable Scalar Network Analyzer (SNA) capable of detecting ice using microwave resonant sensors. The device measures shifts in resonant frequency from Split Ring Resonators (SRRs), allowing ice formation to be detected in real time. The objective was to create a lightweight, field deployable alternative to traditional laboratory network analyzers for use in wireless microwave sensor networks.</p>
The project was initially defined through its proposal and later expanded upon by our client Dr. Mohammad H. Zarifi</a>.</p>

Project Proposal</p> </summary>

Problem statement</strong> - What is the Problem that this project is meant to solve?</em></p>

Microwave resonant sensors are low cost and simple to fabricate for many different applications. However, current microwave readout electronics can be expensive, heavy, and implement many features so they are useful in a variety of different applications. This project will focus on readout electronics specialized in measuring the necessary response from microwave ice sensors. This will miniaturize the electronics required for ice sensing to provide more flexibility on use and portability, such as drones for ice sensing automation. This technology can be used to create safer infrastructure and improve safety for maintenance workers.</p> </blockquote> </li>

Background</strong> - Please provide any additional relevant background information that is related to the problem you would like this project to solve.</em></p>

This project involves designing and development of a two-port circuit implemented on a printed circuit board (PCB), capable of characterizing the frequency response of a resonator within a limited frequency range, dynamic range, physical dimensions, and weight. The system will be integrated wirelessly with smart devices or portable electronics for visualization and communication purposes.</p> </blockquote> </li>

Constraints</strong> - What are the Constraints for the problem you are trying to solve?</em></p>

The objective of this project is to design, build, and implement a scalar network analyzer capable of measuring the amplitude frequency response of resonant based microwave sensors, while being low weight and maintaining a small form factor. Since this device will be implemented in ice sensing projects, the device is expected to withstand temperatures below zero; because of this, the device must remain operational and accurate for temperatures as cold as -40℃ or as hot as 60℃ with a frequency range between 1 to 8 GHz, dynamic range of 50 dB or more, and minimum output power of –5 dBm.</p> </blockquote> </li>

Risks</strong> - What are potential Risks that are associated with the Problem this project is attempting to solve?</em></p>

Personal risks associated with this project may involve handling a soldering iron as well as inhaling any toxic fumes that comes with the soldering process. These risks can be mitigated by informing students of standard safe soldering practices. During the development of the PCB, prototyping of the design may lead to E-waste.</p> </blockquote> </li>

Applicable Engineering Disciplines</strong></p>

Electrical, Mechanical, Manufacturing</p> </blockquote> </li> </ul> </details>
Key takeaways of this proposal were the projects objective, device specifications, and constraints. More on this will be duscussed when going over the desing process</a>.</p>
Meet the team</h3>
With a project such as this a capable team was required. The students assigned to this project were,</p>

Michael Bell</strong> - Electrical Engineering</em></p> </li>

Mathew Bishop</strong> - Electrical Engineering</em></p> </li>

Aiden Macson</strong> - Electrical Engineering</em></p> </li>

Radu David</strong> - Computer Engineering</em></p> </li>

Supreet Dhatt</strong> - Electrical Engineering</em></p> </li>

Duncan Rabenstein</strong> - Electrical Engineering</em></p> </li> </ul>
Each of the students have skillsets in electronics, hardware, and software, making for a strong team to tackle our assigned project.</p>
Starting from the left: Supreet, Radu, Mathew, Aiden, Michael, and Dunctan.</figcaption> </figure>
Design Process (The How</em>)</h2>
The team was tasked with designing a scalar network analyzer (SNA) for wireless sensing and ice-detection applications. The specifications we needed to conform to were as follows:</p>

Operational frequency range of 10-18 GHz (Ku band)</strong></li>
Dynamic range of ≥ 50 dB</strong></li>
Power output of -5 dBm</strong></li>
Measurement resolution of 32 sample points</strong> over range</li>
Low power & portable</strong> optimization (battery operated)</li> </ul>
The key challenges were achieving the desired operational frequency range while maintaining moderate dynamic range.</p>
What is an SNA?</h3>
A scalar network analyzer is a device that will output a sinusoid wave that is swept over a defined frequency range, and measures the amplitude response of a device under test (DUT). This response allows users to see the magnitude frequency response of a device, which in our case allows us to measure microwave ice detection sensors.</p>

SNA RF Subsystem Block Diagram</figcaption> </figure>
Proposed Approaches</h3>
To design this SNA the team focused on the devices RF structure. I was personally tasked with developing the RF subsystems in such a way that there were options balancing performance, cost, and complexity.</p>

Design 1</p> </summary>

Simple VCO and Power Detector</figcaption> </figure>
This design utilizes only a voltage controlled oscillator (VCO) and power detector.</p>

Design 2</p> </summary>

Double Conversion Methodology</figcaption> </figure>
This design utilizes a double conversion methodology incorporating frequency mixing and filtering.</p>

Design 3</p> </summary>

Double Conversion with Multipliers</figcaption> </figure>
This design expands on the double conversion methodology by adding frequency multipliers and amplifiers for additional range and flexibility.</p>

Design 4</p> </summary>

Multiplier and Down-conversion</figcaption> </figure>
This design utilizes frequency multiplication at its output and a down-conversion stage at its input to make it flexible yet simple.</p>

Initial Prototype</h3>
Given the timeline constraints of this project, we decided to proceed with the first design, utilizing a single VCO and power detector. We opted to use connectorized RF components to reduce any possible troubleshooting time in the future. The two major RF components we used were the HMC733 VCO</a> by Analog Devices and the ZV47-K44RMS+ Power Detector</a> from Mini-Circuits.</p>
With these building blocks, the team began working on the device. We used an ESP32 microcontroller in addition to a number of components for power distribution and amplification. A printed circuit board (PCB) was developed to host all the power regulators and voltage conversion stages. The resulting unit was not glamorous, but it demonstrated the projects feasibility.</p>

SNA Prototype 1</figcaption> </figure>
In addition to the VCO and power detector, we used two Mini-Circuits ZX60-R5183P+</a> amplifiers (approx. 8 dB gain). These amplifiers were used for wireless measurements using horn antennas.</p>
Improvements to the software</h3>
With the first prototype built, a simple graphical user interface (GUI) was designed. There were a number of iterations that this software went through. The members working on the software were Supreet, Duncan, and Aiden.</p>

First UI Revision</p> </summary>
The first GUI designed was barebones and made to function-test the first prototype.</p>

GUI Version 1</figcaption> </figure> </details>

Second UI Revision</p> </summary>
The second GUI was a polished version of the first, with more attention to the design and user experience.</p>

GUI Version 2</figcaption> </figure> </details>

Third UI Revision</p> </summary>
The final GUI incorporated all the features the team set out to include.</p>

GUI Version 3</figcaption> </figure> </details>
Final Design</h2>
With the results we were achieving, we looked into areas of our design we could improve. Two distinct areas of improvement we discovered were power consumption</strong> and user experience</strong>.</p>
New PCB</h3>
A new PCB was designed for the device to incorporate new power distribution networks. I was tasked with designing the second PCB iteration which meant designing a new power supply for the device.</p>
We wanted to be able to intelligently control the devices power distribution to optimize for power consumption. To do so, I incorporated a switch-mode power supply to provide the ESP32 with 3.3V, and a switch-mode power supply to provide the 18V required for tuning the VCO. To power the RF components such as the power detector, amplifiers, and VCO, I utilized linear regulators with an Enable/Disable feature for a smooth DC output that I can turn on and off using the ESP32.</p>

Second iteration PCB</figcaption> </figure>
Enclosure</h3>
We achieved an improved user experience for the final design's software as shown in UI Iteration 3</a>, but we also worked to improve the devices presentation as well. An enclosure was designed by Mathew, Aiden, and Radu, to incorporate all the components while keeping the dimensions within the desired volume.</p>

Final Enclosure with Components</figcaption> </figure>
Resulting Product (The Outcome</em>)</h2>
Closing in on the projects completion date the team conducted a final assembly of the device. The result was a product of everyones commitment, hard work, and ingenuity.</p>

Complete SNA Assembled</figcaption> </figure>

SNA Open Case Short Clip</figcaption> </figure>
Testing</h3>
Testing the device, results showed minimal error and useable accuracy, making this device suitable for microwave sensor technologies. Below is a figure of a test done using a SRR. The blue trace shows the SRR's characteristics on a Keysight N5222B vector network analyzer which will be used as our baseline. The red trace is our SNA's measurement. Overall, the relative error between the SNA's results and the baseline was -4.34% in amplitude and +0.35% in frequency.</p>

SRR Measurement Results</figcaption> </figure>
Conclusion</h2>
We successfully built a portable device that allows for microwave ice detection and sensing technology to be implemented at a lower cost. Instead of using large and expensive lab equipment, this device can be deployed in real environments like wind turbines, drones, or aircraft surfaces for ice detection applications that will provide improve safety.</p>
Final Video Presentation</h3>
The final video presentation of our device:</p> </iframe>

Doublet Antenna Design

2025-12-03T00:00:00+00:00

For my antennas course (ENGR 473) in my 4th year of undergraduate studies, the class was given a final project. In this project students were tasked with designing a specific antenna within Cadence's AWR Microwave Office's AXIEM software (I will refer to it as AWR AXIEM, or AXIEM). Students were given a list of antennas they could choose from, such as Yagi-Uda, Bow-tie, Fractal, Helical, etc.</p>

Around the middle of the school term, I was in the process of designing a doublet antenna in the backyard for my HF radio. After a discussion with the professor, he offered I expand on my existing 40m doublet for the project.</p>

The project itself was to hold a presentation educating the class on the history and applications of the antenna, and showing the AWR AXIEM simulation results. In this project writeup I will cover both the process I took of building this antenna and some simulation results.</p>

Doublet antenna on a 30 ft mast</figcaption> </figure>

Doublet Antenna</h1>
A doublet is a common variant of the standard dipole antenna, but with a few key differences. Most notably, a doublet is a non-resonant</u> dipole that uses a balanced feedline (typically a 300 Ω, 450 Ω, or 600 Ω two-wire "ladder" line). While a standard resonant dipole has an impedance of 73 + j42.5 Ω and easily matches a 50 Ω system with a 1:1 current balun, a doublet's feedline impedance varies wildly depending on the frequency. Because of this, it requires an Antenna Tuning Unit (ATU) to achieve a proper match.</p>
Why use a doublet?</strong> It is an incredibly efficient multi-band antenna. Its main draw is the ability to tune across several different frequency bands while maintaining high performance.</p>
What does "non-resonant" mean?</strong> In this context, it means the antenna isn't operating at a natural 50 Ω match. The elements are intentionally cut to lengths that do not match the standard wavelength fractions (like λ/2, λ/4, or λ/8) of the frequencies being used.</p>
Note: when referring to λ, it is the wavelength of the lowest operating frequency</u>.</p>

Operation</h2>
Because doublets are non-resonant, the VSWR on the elements and balanced feedline can reach 10:1 or higher. If you were to use conventional coaxial cable, cable loss would cause these standing waves to dissipate rather than fully radiate. A two-wire feedline solves this problem: it allows the standing waves to reflect back and forth between the elements and the ATU with very little loss, ensuring most of the energy is radiated.</p>
This balanced feedline also acts as an impedance transformer, a detail that influences how a doublet may be built. By intentionally balancing the lengths of the elements and the feedline, the antenna can be targeted for specific amateur radio bands (most noteably, the 80m/75m, 60m, 40m, 30m, 20m, 17m, 15m, 12m, and 10m bands).</p>
Another approach is to cut the elements to resonate on the lowest desired frequency, while relying on the ATU to tune it for higher bands. For the antenna I built and simulated, I used an element length of 10.7m (approx. 44 feet) to establish a minimum operating frequency of 7 MHz for the 40m band.</p>
Building the Doublet</h1>
The antenna was built as affordably as possible. The materials I used to make the antenna were as follows:</p>

6x Fiberglass Military tent poles (OD of 2.75-inch) that I found for CA$5 each at Princess Auto.</li>
A lot of 16 AWG hookup wire that I pulled from a spool I already have.</li>
3D printed components, some crimp connectors, and a small fiberglass panel.</li> </ul>
Constructing the feedline</h2>
Being budget oriented, I decided to construct the feedline myself. To do this, I first calculated the spacing I would need to acieve a 450 Ω characteristic impedance. Using an online two-wire transmission line calculator</a> I determined the best wire spacing to be 28 mm center-to-center. Taking into account the velocity factor of the hookup wire and its PVC insulation, the impedance was likely not exactly 450 Ω, but close.</p>
I designed and 3D printed some spacers that snap onto the wire out of PETG plastic. By placing spacers every 5 to 6 inches along the wires I created the feedline.</p>

Spacer CAD model</figcaption> </figure>
The final feedline was very consistent in its construction and looked the part.</p>

Two-wire Feedline</figcaption> </figure>
Measuring the elements</h2>
The length of each element was cut to 10.7m. This length produces a resonance on the 40m band like a regular dipole antenna, but the doublet configuration allows it to be tunable to multiple bands.</p>

Electrical Connections</figcaption> </figure>
Erecting the Antenna</h2>
With everything measured and cut, the components were ready to be assembled. I constructed the antenna on the hill behind my house, mounting the fiberglass poles inside a secure steel pipe that I had set into concrete.</p>

Doublet antenna components</figcaption> </figure>
The antenna used 44.25-inch fiberglass military tent poles for its mast. I initially used 8 poles to achieve a height of 10m, but it snapped after an exceptionally windy day. Because of this, I shortened the mast to 6 poles, leaving me with a height of 7.8m. This simply meant the antenna had a higher take-off angle, making it better optimized for NVIS operation.</p>

Mast failure</figcaption> </figure>
After shortening the mast and optimizing my guy-line placement, the antenna stands strong.</p>

Doublet antenna standing tall</figcaption> </figure>
Measurements</h2>
To analyze the antenna's impedance over the 1–30 MHz HF frequency range, I used a 1:1 current balun and a NanoVNA to capture the S11 parameters. I then converted the S11 parameters to a Z11 plot and VSWR plot.</p>

NanoVNA measurement setup</figcaption> </figure>
I placed markers on the 80m, 40m, 30m, 20m, 15m, and 10m bands. The resulting impedance at each point will be fed into the ATU via either a 1:1 or 4:1 balun to bring it within a matchable range for a 50 Ω system.</p>

Measured Z11 parameters</figcaption> </figure>

Measured VSWR</figcaption> </figure>
Simulating the Doublet</h1>
The doublet was simulated in AWR AXIEM, however due to constraints in the software, I had to scale the antenna and thus its frequency of operation. I scaled the frequency by 100 (and lengths scaled by 0.01). The simulations were run from 500 MHz to 3000 MHz.</p>
I simulated different parameters of the antenna:</p>

Total feedline length</li>
Height off ground</li>
Different earth conductivities</li> </ul>
Feedline Length</h2>
Since the feedline acts as an impedance transformer, certain configurations can create extreme impedances that are challenging for an ATU to handle. Specifically, it is best to avoid a feedline length that is an integer multiple of λ/2, or any combination where the feedline's electrical length plus one antenna element equals an odd multiple of λ/8 (such as λ/4 + λ/8 = 3λ/8).</p>
</figure>
The simulation evaluated the resulting impedances for three different feedline lengths: 107.1 mm, 160.7 mm, and 190 mm. The 190 mm feedline yielded the best results, as its non-multiple length provided the most matchable impedances.</p>
</figure>
Antenna Height</h2>
Because the ground mirrors the currents in a horizontally oriented dipole or doublet, the ground plays an integral part in the far-field radiation pattern of the antenna. The best performance is acheived by keeping the antenna λ/2 from the ground or higher. When the antenna elements are less than λ/2 off the ground, the mirrored currents in the ground will act against the antennas currents providing destructive interference. This interference pushes the take-off angle upwards, making the antenna radiate towards the sky rather than the horizon. Having the antenna less than λ/2 or even λ/4 is the principle behind near vertical incidence skywave (NVIS) operation.</p>
The resulting radiation patterns show that as the height off the ground decreases, the main radiating lobe of the lowest frequency is directed further upwards.</p>
</figure>
Ground Conductivity</h2>
Because the ground interacts with the antenna's radiation pattern, I simulated the effect of different earth conductivities based on One-dimensional Layered Earth Models</a> provided by the canadian geological survey.</p>
</figure>
Presentation Document</h2>
The full slide deck can be found in PDF format here:

Final Design</h2>
With the results we were achieving, we looked into areas of our design we could improve. Two distinct areas of improvement we discovered were power consumption</strong> and user experience</strong>.</p>

Final Video Presentation</h3>
The final video presentation of our device:</p> </iframe>

Presentation Document</h2>
The full slide deck can be found in PDF format here:

Final PCB</h2>

</video>
Finished PCB</figcaption> </figure>

Michael Bell

Superheterodyne HF Receiver

Revisiting the Cascode Amplifier

Undergraduate Capstone Project

15 GHz Split Ring Resonator

Homebrew Ribbon Microphone

Inductor Saturation Current Tester

Vacuum Flourescent Display Filament Driver

Doublet Antenna Design

2025 UBCO IEEE PCB Design Workshop

RF Mixer Visualization

Michael Bell

Superheterodyne HF Receiver

Revisiting the Cascode Amplifier

Undergraduate Capstone Project

Final Design</h2> With the results we were achieving, we looked into areas of our design we could improve. Two distinct areas of improvement we discovered were power consumption</strong> and user experience</strong>.</p>

Final Video Presentation</h3> The final video presentation of our device:</p> </iframe>

15 GHz Split Ring Resonator

Homebrew Ribbon Microphone

Inductor Saturation Current Tester

Vacuum Flourescent Display Filament Driver

Doublet Antenna Design

Building the Doublet</h1> The antenna was built as affordably as possible. The materials I used to make the antenna were as follows:</p>

Measuring the elements</h2> The length of each element was cut to 10.7m. This length produces a resonance on the 40m band like a regular dipole antenna, but the doublet configuration allows it to be tunable to multiple bands.</p>

Erecting the Antenna</h2> With everything measured and cut, the components were ready to be assembled. I constructed the antenna on the hill behind my house, mounting the fiberglass poles inside a secure steel pipe that I had set into concrete.</p>

Presentation Document</h2> The full slide deck can be found in PDF format here: </svg></span> Presentation Slides</a>.</p>

2025 UBCO IEEE PCB Design Workshop

Final PCB</h2> </video> Finished PCB</figcaption> </figure>

RF Mixer Visualization

Final Design</h2>
With the results we were achieving, we looked into areas of our design we could improve. Two distinct areas of improvement we discovered were power consumption</strong> and user experience</strong>.</p>

Final Video Presentation</h3>
The final video presentation of our device:</p> </iframe>

Presentation Document</h2>
The full slide deck can be found in PDF format here:

Final PCB</h2>

</video>
Finished PCB</figcaption> </figure>