Vertigo - Matthew Warren

Vertigo was designed as a long range VTOL UAV. The design process started with the mission definition. I wanted a:

Cruise speed of 40 mph
Cruise endurance of 1 hour
Hover endurance of 5 min

Quadplanes are mechanically simple because they have no tilting or moving motors. There are 5 motors on a quadplane, 4 for takeoff and 1 for forward flight. This configuration is not optimal because there are unused motors in both hovering or forward flight. However, there are efficiency advantages to having dedicated propulsion systems. No compromise needs to be made to hover and cruise with the same components. I chose to build a blended wing body to reduce drag and because all of the internal components are fairly thin.

I used AVL to size the aerodynamic surfaces and excel to choose the scaling and propulsion system. An inverted V-tail was used both for structural convenience and because it allowed for a single pusher motor that doesn't place the tail in the propeller wash. The large, inverted tail required some polyhedral to ensure proper directional stability. An angle of 80 degrees was chosen for the V-Tail to balance the longitudinal and lateral stability.

I created regressions for motor and battery weights as a function of thrust and capacity. I also had estimates for structural weight fractions and required control system weight. Using these parameters, I could size the airplane to meet the performance requirements. I then looked for motors near the predicted motor weight with the required thrust and power based on the performance requirements.

The most efficient motors for hovering generally have a low power to weight ratio. As the motor power to weight increases, generally the efficiency decreases. Heavier motors are of course worse in cruise. In addition, larger motors with larger rotors are more efficient but those rotors produce more drag in cruise mode. I chose the t-motor MN4010 motor rated for 400 W spinning 16 inch propellers for hover and a cobra 2820/14 rated for 550 W for the pusher motor spinning an 11x6. The airplane weighed 10 lb at takeoff and carried 16 Ah of 4s LiPo batteries. This combination represented a good trade-off between hover power required and hover propulsion system weight. I ended up biasing toward hover efficiency despite the cruise efficiency loss to reduce the load on the batteries in hover and to increase the hover time for testing.

The structure needed to be transportable in the back of my sedan. The main loads pass through a carbon wing spar and two carbon booms. The tail has a carbon spar running through each half. Highly loaded sections of the structure use plywood such as the motor mount, fuselage core, and the top of the tail where the carbon rods join. Thin 3D printed fairings fit behind the rectangular boxes used to attach the motors to the carbon fiber booms. The wing tips are removable using ferrules, the tail booms are removable with bolts in the wing, and the tail is removable using bolts going through the tail booms. The wood structure was laser cut with attached jigs that made assembly quick and precise. The structure was very light. The tail was lighter than predicted and actually needed a small amount of weight to keep the CG in the correct location.

While the structure ended up being slightly under weight, the electronics were slightly over. I underestimated the volume of the wiring and the overall bulk of the wires. Wire management became a struggle, and I ended up wishing I had left larger channels for the main power cables and connectors in the center section near the batteries. Despite the drag penalty, the vertical lift ESCs were mounted externally for cooling.

A Pixhawk running PX4 was used for the flight control system. I needed to mount the GPS slightly elevated and on an aluminum sheet to prevent loss of signal. (This will be important later) I tuned the quadcopter controller first. It is very important to have the quadcopter PIDs set well before attempting a transition. If they are not set properly, the airplane can become unstable in the transition once the back pusher motor is activated. The airplane controller was tuned once the airplane could be transitioned smoothly into manual control. After adjusting all the gains, the airplane could takeoff autonomously, transition to cruise, fly around, return back to hover, and land with no human interaction.

I mounted all the motors vertically but would suggest tilting them out by about 7 degrees to improve yaw performance. My airplane relied on differential torque to yaw. Tilting the motors out adds a thrust component to the torque effects greatly improving the yaw performance. My aircraft had most of its weight near the CG but if the weight was further out toward the tips, it would have struggled to yaw. There is an essentially no penalty to tilting the motors out; a 1% loss at 7 degrees.

Unfortunately, Vertigo is no longer with us. There was an issue during a fully autonomous test flight with the GPS. The airplane was in cruise when the GPS cut out. The airplane followed proper protocol and transitioned to hover and attempted to land. The pilot (I) was not aware there was an issue but noticed the airplane descending and drifting. I placed the airplane back into manual AIRPLANE control and attempted to turn. From my perspective I couldn't tell the airplane had slowed down significantly in hover mode. I thought it was still flying as an airplane, just in the wrong direction. My command to turn in airplane mode led to an asymmetric tip stall that destroyed the aircraft.

I believe this aircraft is one of the best I have ever designed. It had great flying characteristics and performance. While it is tempting to rebuild, the time required made this impossible without abandoning other ongoing projects. I am looking at ways to make a simpler version with equally good performance.