The electronics comprise the various systems of the Hoverfish, with the wiring providing power and signals to each of the actuators and the various servos providing the forces for the fish to move.

Wiring

Simple and often overlooked, the wiring on the Hoverfish originally amounted to a sizable percentage of the total weight on board. At the start, we chose to use thicker single core wires as we believed that the wires would cause a large amount of resistance, potentially damaging the balloon and draining the battery quickly. While the wires served their purpose well, they were heavy and stiff, qualities we sought to replace.

Upon receiving our first batch of servos, we realised how thin the accompanying wires were. We thus concluded that the wires mounted on the Pi were excessively thick and sought to replace them. Unfortunately, the only thinner replacement were mulitcore wires from the MnT lab, which came in only 2 colours. Furthermore, the mutlicore wires were not as stiff, making it difficult to interface with the servos. Both issues were solved by attaching the wires to a 2.5mm header, allowing us not only to easily plug the wires in and out of different servos, but also allowing us to mark out the correct orientation relative to the servo with only 2 colours of wire. Ground and live used red wires, whereas the signal used a green wire.

Soldering the mutlicore wires was no easy task as they often frayed apart. We also had to resolder them mutliple times as they would break easily. At the end of the day however, they still serve their purpose, despite our mediocre soldering skills.

Power

Energy is the source of all life. Without a suitable energy source the Hoverfish is not more than a floating fish, dead and devoid of life. Through our various tests, we realised that choosing a appropireate power source was vital to grow the spark of life into a blazing fire. Given that the various components were powered from the Pi, we needed a consistent 5V power supply to supply the Pi. To achieve that, we used a DC/DC boost to increase the voltage provided by a Lithium Polymer battery from 3.7V to the 5V required. The DC/DC boost was connected to the Pi through our wires as well as a repurposed micro USB head. Performing some run time tests and weight considerations, we settled on a 500MaH LiPo weighing in at 9g and providing 30mins of uptime.

From there, we had two choices of DC/DC boosters, the adafruit Powerboost 500 and 1000, providing diffferent amounts of power out. Originally we had started with the lighter Powerboost 500 which was sufficient for our actuators at the start. As the Hoverfish came closer to completion, we realised that the various components were drawing more power than the 500 could provide, resulting in constant reboots of the Pi. As such, we stepped up to the 1000 to meet our power needs. There came to a point however, when even the 1000 was insufficient, leading to us having to step back on our servo.

In conclusion, the our power supply created yet another limitaion we had to balance.

Servos

The limbs of the Hoverfish, the options we considered here fell into 3 catergories: ultra-micro, 6g micro and 9g micro servos. The ultramicro servos were extremely light, weighing roughly 3g, with torques of about 100g/cm, taking 3.3V. The 6g microservos had torques of about 700g/cm while the 9g microservos had torques of about 1.5kg/cm at around 5V.

In the beginning while we were obsessed about weight, we chose to go for the ultramicro servos. However we were met with two issues with this. The first is we used the 5V power pins from the Pi, over loading the servos. The second is that we underestimated the torque required both to flap the tail and move the position of the CG.

We then tested a 6g micro servo and found the performance to be satisfactory, subsequently leading us to use them for both actuators. While this worked for a while, we soon discoved that we had fried the servo used to control the position of the CG, leading us to chose to opt for a larger 9g microservo.

Upon implementing the 9g microservo, we were met with many power issues, with the Hoverfish powered by the Powerboost 1000 facing frequent disconnents before the Powerboost 1000 burnt out, and the Powerboost 500 rebooting whenever the 9g microservo was used.

Finally, we returned to the 6g microservos, cutting the torque requirements to meet the power requirements.

Capacitor

To allow our power supply to accommodate stronger servos than it would usually be able to support, a capacitor was added to the buoyancy servo. Its purpose was to store charge to avoid sudden power spikes to the servo causing the Pi to reboot.