Overview

In this project, a compact, deployable communications architecture was created for use with CubeSats as small as 1U for the CubeSat Flight Laboratory (CFL) at the University of Michigan. The system was to replace existing communications infrastructure already in use by CFL, which suffered from poor link reliability due to antenna material and design.

The prior design utilized a linearly polarized stainless steel flat monopole array. This configuration imposed directionality constraints; when spacecraft rotation caused a null in the antenna radiation pattern (inherent to this architecture) to face the ground station, the communication link was lost.

I led a team tasked with developing a more robust communications system that mitigated these limitations while incorporating additional functionality requested by CFL. My responsibilities included RF implementation and testing, all electrical hardware design, and conducting all research and trade studies relating to the communications architecture.

The system was designed to be fully space-rated, and was fully completed with a prototype delivered to CFL. This frontend is planned to be flown on upcoming CFL missions in Low-Earth Orbit.

Key Features

• 1U CubeSat-compatible footprint
• LoRa-based architecture (UHF Band)
• Nichrome burn-wire deployable quad-monopole system
• Passive and bidirectional hybrid coupler integration
• Innate circular polarization

---

Design Notes

---

Initial Requirements Breakdown

After a meeting with the client, a NASA-style requirements breakdown was performed, decomposing the system into a list of customer needs, key system drivers, constraints, and functional requirements.

Through this process, key parameters such as the LoRa spreading factor, data rate, antenna material, and form factor were determined. One of the key constraints was the physical realization of a CubeSat-sized system which can operate at the requested UHF frequency band of 433 - 437 MHz. The resonant wavelength for this band is approximately 68 cm, meaning that a quarter-wave monopole antenna length for this band must be approximately 17 cm. As the system was designed to fit within a 1U cubesat form factor (a cube with side lengths of 10 cm), antenna design, deployment, and RF signal conditioning presented a significant challenge compared to a system of larger size or higher frequency.

Antenna Radiation Pattern Simulation

As one of the driving system requirements was omnidirectional, orientation-independent communication, radiation patterns were simulated in MATLAB to allow for both qualitative and quantitative refinement. Below are two examples taken from these simulations during the optimization process.

Numerical optimization revealed that a polar inclination of approximately 30° yielded an optimal balance between antenna signal power and omnidirectionality.

Polarization and Hybrid Couplers

One key improvement requested by CFL over the predecessor system was the inclusion of circular polarization. Integrating circular polarization provides two key benefits: orientation independence and Faraday rotation immunity. For a pair of linearly polarized antennas, the received signal power depends upon the relative angular orientation angle \( \theta \) between the transmitting and receiving antenna, represented as

\[ P_r = P_t \cos^2(\theta). \]

As such, signal loss due to orientation misalignment approaches 100% as \( \theta \) approaches 90°. Implementing circular polarization eliminates this dependence as a circularly polarized field consists of two orthogonal linear components 90° out of phase. Any linear orientation couples to one of these two components. Functionally, this means that in the worst-case scenario, an orientation mismatch now results in at most a ~3  dB signal loss rather than a total communications blackout. This increases communications uptime and prevents other spacecraft systems (such as ADCS) from having to carefully manage orientation to facilitate basic communication.

In addition, circular polarization mostly negates an ionospheric phenomenon known as Faraday Rotation. As signals pass through the ionosphere, the polarization vector rotates due to interaction with charged particles captive within Earth's magnetic field. This rotation angle \( \phi_f \) scales in relation to the total electron content of ionized plasma in the transmission path divided by the square of the transmission frequency. For UHF in typical ionospheric conditions, this can result in a rotation in the range of 20° - 120°. For linear systems, this rotation acts the same as if one antenna was physically misaligned by the same magnitude of angle. For circular systems, any rotation caused by Faraday effects is mostly nullified, possibly causing slight elliptical polarization behavior in the worst case.

To realize circularization, ideally a branchline coupler would be realised using a microstrip topology directly on the circuit board, using an impedance-controlled substrate. Unfortunately, these types of coupler also possess a requirement of a side length of at least \( \lambda\)/4, which would require a board size larger than our form factor allowed. There are ways around this (meandering labryinthine traces, abusing multiple board layers, exploiting strong dielectrics, etc), however these all result in nonlinearities and other departures from the ideal textbook behavior.

As such, prepackaged hybrid couplers from Mini-Circuits were implemented to achieve the desired polarization. A 180-degree ferrite core and wire hybrid (Model SYPJ-2-13) provided an initial 180° split, outputting two lines at half power (~3 dB) and 0/180° phase shift from the source signal. Two QBA-07+ quadrature couplers further split these two phases into a total of four evenly spaced 90° phases at a nominal 0°, 90°, 180°, and 270° from the original signal. These quadrature couplers utilized Low Temperature Co-Fired Ceramic (LTCC) technology, which realizes a lumped model LC network directly imprinted within the device. Electrical trace impedance and length matching was performed as seen in the below PCB layout to ensure that the desired 90° phase shift was preserved throughout the entire component chain from input, through the couplers, to the antenna interfaces.

The system was paired with a Semtech SX1276 transceiver capable of switching between TX and RX mode without any external circuitry. As the packaged couplers are fully bidirectional as well, when in receive mode, this RF frontend can recombine an incoming signal of matching polarity back into a single signal at the feedline interface. This system was designed to be a minimally-complex, low cost solution, and as such is intended to run in half-duplex mode, however with the addition of an additional transceiver, the coupler topology could easily be adapted to realize a crude orthomode transducer, allowing simultaneous TX/RX with opposing right and left-hand circular polarizations.

Images

Mechanical Antenna/Deployment Design

The complete mechanical design was delegated to a separate mechanical engineering team, however the antenna design, material, and deployment concept was within the scope of the initial prototype deliverables. As mentioned previously, each monopole must have a length of at least \( \lambda\)/4, which in this case is 17 cm. To fit within the 10 cm side length constraint, a hollow, pivoting design with a midpoint hinge was devised. The antenna exterior consists of a space-rated, RF safe fiberglass tube, invisible to radio waves at this frequency. Inside the tube a beryllium-copper antenna wire, cut to length equal to the midpoint of the desired bandwidth's wavelength is affixed.

The antenna is affixed to the board via a rotating heim joint, allowing rotation in both the azimuthal and polar directions. The tube is separated approximately halfway up its length, and folded parallel with a side-mounted traditional hinge. Two springloaded mechanisms are present in each antenna, one near the base heim joint and one at the midpoint hinge, and are constrained by redundant nichrome burn-wires pre deployment. When the system is deployed, the first burn-wire releases the folded mechanism from its stowed position within the board, and a machined ramp and stop guides each antenna to a 90° rotation in azimuth and a 30 degree rotation in polar angle. Once the first deployment step is completed, the second set of burn wires releases the midpoint hinge, allowing the antennas to extend to full length.

Note that the antennas used in the initial prototype are COTS monopoles, which are less than 17 cm but still possess a resonant frequency near 437 MHz.

Testing and Results

The completed prototype was tested both within lab settings as well as in an outside analogue to a real-world setting. A calibrated Vector Network Analyser (VNA) was first used to examine the S₂₁ response between the signal input and each phase, as well as the relationship between phases. As seen below, the system was able to maintain orthogonality between each phase within a margin of less than 5° meeting client requirements. This nonideality translates to a delay mismatch of only ~25  picoseconds, likely caused by tolerances in the hybrid couplers or meandering trace realization.

In addition to the nearly perfect phase separation, the power split between each phase was near-ideal. In a perfect system, each phase is attenuated 6 dB relative to the input signal; with the 180° coupler halving the source (into two -3  dB signals) and each quadrature coupler halving those signals again. All phases exhibited between -6.5 and -8  dB of attenuation, representing a worst-case difference of ~15% amplitude between the lowest and highest phase, causing slight elliptical behavior. This difference was likely due to losses in the 180° ferrite hybrid coupler, as the lower paired attenuation was offset by 180°.

Physical parameters such as SNR and RSSI were verified in University of Michigan's anechoic chamber, and long-range reliability was verified by testing pairs of the full system successfully at multi-km distances. RSSI varied depending on the context, with received signals ranging anywhere between -30 and -100 dBm using a broadcast strength of +20  dBm. These RSSI numbers should be taken with a grain of salt however, and were mostly interpreted as qualitative system verification, as ground reflections and multipath interference greatly impeded signal transmission compared to the free-air scenario that this system is designed for. Ideally, further testing on a high-altitude balloon would be performed to validate long-range communication performance before this system is launched as part of a CubeSat.