Finished version v1.0

This is the final version 1.0 with a 2 axis gimbal build with Meccano (which didn't last for more than 2 flights).

First test flight

This was it... We took the quad to an open space and finally did a real flight test. No strings, no holding it down, nothing... And it worked!!!!

Here's the complete configuration of this working version (v1.0):

Arduino MEGA board

Y83 IMU 9-DOF (or similar)

4x Turnigy 2830-800Kv  motors

4x ESC Hobbyking SS series HK-40A

4x Props 10x4.5

Turnigy 3600 mAh LiPO battery

Futaba T14SG 2.4GHz radio

Firmware: MultiWii v2.0

The vibration problem

If you want a steady quadcopter (or multi-copter), you need an accurate reading of angles (pitch and roll).

The reading is made through a combination of calculous (which I find really hard to understand) that mixes the information of a 3-axis accelerometer, 3-axis gyroscope and a 3-axis magnetometer.
When you have 4 motors spinning at 10,000 RPM each anywhere near your inertial sensor, you will probably sense all the vibration. I can bet you will.

When the vibration is very hard, no software correction or filter is possible. So a mechanical solution is needed.

This is how we solved the problem:

IMU anti-vibration mount

The IMU (9 DOF) is attached on an independent board, which is "floating" on rubber mounts. This absorbs most of the vibration generated by the motors, and the rest of the vibration is filtered out by software.

First 4-motor test

The four motors will lowly increase RPM until a lift is detected with the sonar located on the bottom. Then, everything will stop running.

The frame (version 0.1)

CAD model

First frame layout finished

Building motor mounts

Finished motor mount

Finished motor mount 

All motors and props in place :-)

First PID tests

From Wikipedia: A proportional-integral-derivative controller (PID controller) is a control loop feedback mechanism (controller) widely used inindustrial control systems (Programmable Logic Controllers, SCADA systems, RTU's etc). A PID controller calculates an "error" value as the difference between a measured process variable and a desired setpoint. The controller attempts to minimize the error in outputs by adjusting the process control inputs.

A quadcopter uses a PID mechanism to stabilize itself. It measures the inclination angle with an IMU (Inertial measurement unit) and the corrects the error applying power to each motor.

This is our first low budget test we made with a PID model using a Servo,  a sonar to measure distance, a potentiometer to establish the desired set point, and a piece of paper :-)

Then, it got more complex. Same test but using a real motor, ESC and IMU:

In order to do this test, we had to create a plotting library for Processing. You can download it for free.