Quaffle

2012/01 -- 2012/04

===

# Introduction

_Quaffle_ is a quadrotor flying robot. It uses an [Arduino Mega 2560](http://arduino.cc/en/Main/ArduinoBoardMega2560) processor. The inertial measurement unit is an SPA 9DoF chip. The chassis is completely custom-designed and is constructed from carbon-fibre tubes, 3D printed ABS thermoplastic, and a small amount of sheet metal cut using a waterjet machine.

The robot was built as a team of three: [Anson Liang](http://www.ansonliang.com), [Richard Lee](http://richardxlee.com), and myself. The project was for the ENPH 459 (Engineering Physics Project I) course work.

pic quaffle1.jpg Quaffle taking off : Three photos of Quaffle taking off during an indoor test.

The original intention was for fully autonomous flight using a wireless camera mounted beneath the quadrotor, which streams video to an external laptop computer that sends flight commands to the quadrotor. However, due to lack of time, we were unable to achieve this objective. Nonetheless, we have an awesome quadrotor.

# Mechanical design

The scale of the quadrotor is comparable to commercial remote-controlled aircraft, at about 50 cm wide. Using lightweight components such as carbon fibre chassis and lithium polymer battery, the quadrotor weighs 1.3 kg fully loaded.

The chassis of Quaffle was designed to be rigid but lightweight. It consists primarily of carbon fibre tubes connected at the corners by 3D printed plastic parts and at the centre by polycarbonate and aluminium components fabricated using an [OMAX waterjet cutter](http://www.omax.com/). The design emphasizes modularity and scalability as well as low cost and ease of fabrication and assembly. The size of the quadrotor can be easily changed by using carbon fibre tubes of different lengths,
and the central assembly can be expanded to install additional instruments by simply adding more layers on top.

pic pics/Quaffle.jpg Quaffle : Photograph of complete Quaffle prototype.

## Carbon fibre tubes

The four arms of Quaffle are carbon fibre tubes, selected because of their strength and light weight.
Quaffle uses two different types of circular carbon fibre tubes: the thicker type is used for the main quadrotor arms, and thin auxiliary tubes connect the arms to each other.
The four arms are cellophane-wrapped unidirectional carbon fibre tubes with an outer diameter of 14.04 mm and an inner diameter of 12.45 mm.
In between each pair of two adjacent arms, a thinner unidirectional carbon fibre tube of diameter 4.76 mm connects them near the end.
The purpose of these tubes is twofold: They ensure the motors point upwards (which would otherwise be difficult due to the circular tube), and reduce vibration and improve rigidity.

pic pics/quaffle.svg Schematic of chassis : Schematic of Quaffle chassis not including motors, rotors, and electrical components.
pic pics/cftubes.png Carbon fibre tubes in Quaffle : Diagram indicating the position of the carbon fibre tubes on the Quaffle chassis. Highlighted in red are the main arms (left) and the auxiliary tubes (right).
pic pics/cftubestudy.png FEA simulation results of carbon fibre tubes : Finite element analysis results of the carbon fibre tube used in the main arm. For an applied force of 10 N, the maximum deformation is about 0.9 mm with a maximum von Mises stress of 29 MPa. The deformation scale shown here is approximately 32.68.

| Property | Value |
| Density | 1.60 g/cc |
| Young's modulus (axial) | 135 GPa |
| Young's modulus (transverse) | 1 GPa |
| Ultimate tensile strength (axial) | 1500 MPa |
| Ultimate compressive strength (axial) | 1200 MPa |
| Ultimate tensile strength (transverse) | 50 MPa |
| Ultimate compressive strength (transverse) | 250 MPa |
Mechanical properties of unidirectional carbon fibre tubes.

## 3D-printed components

Quaffle features several parts fabricated using a 3D printer. 

### Corners

We used the [PP3DP](http://pp3dp.com) in the Engineering Physics Project Lab to create the "corners", which connect the arms to the motor mount as well as to the auxiliary tubes.

pic pics/quadrotorcorners.png 3D printed corners : Diagram indicating the position of 3D printed corner pieces on the Quaffle chassis, highlighted in red.
pic pics/3dprinting.jpg 3D printing the corners : Photograph of three quadrotor corners being printed.
pic pics/quadrotorcorner.svg Schematic of corners : Schematic of quadrotor corners.

### Instrument mounts

The battery holder and the rangefinder mount were both 3D printed.

pic pics/holdingthings.png Battery and rangefinder mounts : Diagram indicating the position of 3D printed battery holders (left) and the rangefinder holder (right) on the Quaffle chassis, highlighted in red.

## Polycarbonate components

Polycarbonate parts were used in both the central assembly for mounting electronic circuits, as well as on the corners as landing gear. We used 3--6 mm thick polycarbonate cut using the waterjet cutter.

pic pics/lexan.png Polycarbonate components : Diagram indicating the position of Arduino mounting platform (left) and polycarbonate portion of central assembly (right) on the Quaffle chassis, highlighted in red.
pic pics/clamps.png Render of central assembly : Render showing close-up exploded view of central assembly. The three polycarbonate layers are clearly shown, as well as the semicircular clamps used to secure the carbon fibre tubes. In the center is an aluminium core.
pic pics/feet.png Quaffle's feet : Diagram indicating the position of the landing gear of the Quaffle chassis, highlighted in red.

## Metal components

The use of metal was kept to a minimum to minimise weight, except in places where the strength and rigidity of metal is needed. These include the motor mounts (sheet steel) and the central core (aluminium), both cut using the waterjet cutter.

pic pics/metal.png Metal components : Diagram indicating the position of the central aluminium piece (left) and the steel motor mounts (right) on the Quaffle chassis, highlighted in red.
pic pics/motormountstudy.png FEA result of motor mount : Finite element analysis result of motor mount under axial load. For an applied load of 10 N, the deformation (left) is around 0.01 mm and the maximum von Mises stress (right) is 22 MPa. The deformation scale here is approximately 352. The material is mild steel.

The motor mounts were cut from 20-gauge (0.75 mm thick) mild steel and the aluminium core was made from 4.8 mm thick aluminium.

# Electronic components

## Arduino Mega 2560 microcontroller

We used the [Arduino Mega 2560 microcontroller](http://arduino.cc/en/Main/ArduinoBoardMega2560) because of the large amount of available software for it. It is a popular platform for hobbyist micro aerial vehicles.

pic pics/arduino.jpg Arduino : Arduino Mega 2560. Image credit [Arduino](http://arduino.cc/en/Main/ArduinoBoardMega2560).

## Inertial measurement unit

The inertial measurement unit in Quaffle is a [SparkFun 9 degrees of freedom IMU](https://www.sparkfun.com/products/10724).

pic pics/imu.jpg IMU : SPA 9DoF. Image credit [SparkFun](https://www.sparkfun.com/products/10724).

This chip interfaces with the Arduino through the I2C protocol.

## AeroQuad shield

The AeroQuad shield v2.1 is used for interfacing with sensors and outputs.

# Electric components

## Motors and electric speed control

We used the Turbojet 880 KV brushless motor. This is a very powerful motor typical of micro aerial vehicles.

pic pics/motor.jpg Motor : Turbojet 880 KV brushless motor (rotor not shown).

| PWM duty cycle | 25% | 50% | 75% | 100% |
| Current (A) | 1.5 | 6.1 | 14.2 | 20 |
| Power (W) | 17.5 | 70 | 160 | 210 |
| Thrust (g) | 230 | 650 | 1290 | 1380 |
Specifications for the Outrunner 880 brushless motor.

pic pics/esc.jpg ESC : Electronic speed controller for the Turbojet 880 KV brushless motor.
pic pics/escdiagram.jpg Wiring schematic for ESC : Wiring schematic for electronic speed controller.

## Wireless communication

Quaffle can communicate with external devices using two methods. There is a 2.4 GHz remote controller to perform manual test flights. This is a HobbyKing remote controller with 7 channels. The other method is a Bluetooth shield for the Arduino.

## Battery

The battery is a typical lithium polymer battery. We used the 3-cell Zippy LiPo 11.1 V battery with 2650 mAh. This allows for 11 minutes of flight when the motors are drawing 15 A (in reality the motors will not be drawing so much current all the time).

pic pics/battery.jpg Battery : Zippy LiPo 11.1 V 3-cell battery. Image credit [Hobbyking](http://www.hobbyking.com/hobbyking/store/__9947__ZIPPY_Flightmax_2650mAh_3S1P_40C.html).

## Power distribution

pic pics/powerdistribution.svg Power distribution schematic : Power distribution schematic. Thicker lines indicate 20 AWG wires.

# Results

We achieved manual indoor flight, but due to lack of time, we did not get autonomous control to work. While it flies alright, we noticed some mechanical deficiencies, such as excessive vibration in the motor mount. The vibration caused a significant amount of noise in the IMU readings, making autonomous control difficult.

pic pics/motormountstudy2.png FEA study of motor mount with transverse forces : Finite element analysis of motor mount with transverse forces.

From finite element analysis, the transverse deformation under a 1 N load is 0.04 mm. Using this, we can find the critical frequency in the transverse direction:

$ \begin{align}
\omega_n &= \sqrt{\frac{k_{eq}}{m_{eq}}}\\
&= \sqrt{\frac{1\text{ N}/0.04\text{ mm}}{110\text{ g}}}\\
&= 477\text{ rad/s}\\
&\approx 4500\text{ rpm}
\end{align}

This is entirely within the operating range of the rotor, so even if the rotor was slightly off-center, it could cause an excessive buildup of vibration.

pic pics/vibrations.jpg Photo of vibration : Photo of rotor during flight with a 1/40 s exposure. The faint white trails of the rotor tips indicates there is some vibration, and that it is around 4800 rpm.

Right now the quadrotor is out of commission because one of my friends took the battery to Toronto and I have never seen him since. So the quadrotor is sitting on a bookshelf in my basement. With commercial drones being really cheap nowadays, it might not be cost efficient to resurrect Quaffle.