Você está na página 1de 79

DREXEL UNIVERSITY

Senior Design Report


Gimbal-Actuated VTOL Nacelle
Team 32
Matthew Brenner, Graham Donaldson, Chris Killian, Henry Tarplin & Robert Walto
Advisor: Dr. Ajmal Yousuff
Submitted May 23rd, 2014

Abstract
There currently exists a large disparity between the maneuvering capabilities of small-scale
UAVs (such as those currently being developed in university research environments) and large-scale
military aircraft. With the exception of the V-22 Osprey, current turbine-propelled fixed-wing military
aircraft are restricted to limited adjustments of their velocity in the body-forward direction and
controlling their vertical position by way of control surfaces (primarily flaps and elevators).
Small-scale UAVs have been demonstrated to have extremely precise control over their
position and velocity in all six degrees of freedom (latitudinal, longitudinal, vertical, roll, pitch and
yaw). However, their current primary propulsion sources are electric motors, which are less energy
dense than turbine engines and jet fuel. This method would therefore not scale well to larger aircraft,
due to current motor designs and the significant size of the batteries that would be required to store
the requisite energy.
As such, there is a need from those stakeholders in military, commercial, and civil fields for a
new propulsion control method that can continue to utilize the high-performance systems currently
in development and production while incorporating the maneuverability and advanced aerial control
currently being developed in small-scale UAVs.
The senior design goal was to design the biaxial gimbal hardware and accompanying control
system, which is independent of the propulsion source, to steer the aircraft and provide vertical takeoff and landing (VTOL) capacity by moving the direction of the thrust vector instead of the
magnitude. The project includes the designing, modeling, fabrication, and testing of a prototype
vehicle featuring three Gimbal-Actuated VTOL Nacelle (GAVN) units in a triangular configuration
as a proof of concept platform. The aim was to prove that these GAVN systems are an effective
method of VTOL aircraft maneuvering.

ii

Contents
Abstract .................................................................................................................................................. ii
List of Figures ...................................................................................................................................... iv
List of Tables ....................................................................................................................................... vi
Introduction .......................................................................................................................................... 1
Background ....................................................................................................................................... 1
Vertical Takeoff and Landing .................................................................................................... 1
Magnitude Controlled Steering .................................................................................................. 2
Direction Controlled Steering .................................................................................................... 3
Stakeholders and Needs .................................................................................................................. 5
Problem Statement........................................................................................................................... 6
Methods ................................................................................................................................................. 6
Project Framework .......................................................................................................................... 6
Requisite Skills and Modern Engineering Tools ......................................................................... 9
Concept Generation............................................................................................................................. 9
Specifications .................................................................................................................................... 9
Concepts ..........................................................................................................................................10
Concept Evaluation .......................................................................................................................13
Design Process....................................................................................................................................15
Theoretical Modeling .....................................................................................................................15
Dynamics.....................................................................................................................................15
Code .............................................................................................................................................19
Detailed Design ..............................................................................................................................20
GAVN .........................................................................................................................................20
Airframe ......................................................................................................................................32
Electrical System ........................................................................................................................36
Prototyping .....................................................................................................................................40
GAVN Component List ...........................................................................................................40
Center of Gravity Testing .........................................................................................................41
3D Printing .................................................................................................................................42
Thrust and Voltage Testing ......................................................................................................44
GAVN Test Stand Design & Fabrication ..............................................................................46
iii

Control System Component Testing ......................................................................................46


Code Development ........................................................................................................................50
Performance Testing......................................................................................................................53
Project Impacts ...................................................................................................................................55
Economic and Social .....................................................................................................................56
Environmental and Ethical ...........................................................................................................56
Project Management ..........................................................................................................................57
Team Organization ........................................................................................................................57
Schedule and Milestones ...............................................................................................................57
Project Budget ................................................................................................................................59
Discussion ...........................................................................................................................................59
Summary & Conclusions ...................................................................................................................61
Future Work ........................................................................................................................................62
Acknowledgements ............................................................................................................................62
References ...........................................................................................................................................63
Appendices ..........................................................................................................................................64
Appendix A: Assorted Images .....................................................................................................64
Appendix B: Thrust Test Data .....................................................................................................70
Appendix C: Tricopter Flight Dynamics Code..........................................................................72

List of Figures
Figure 1: F-35B in VTOL Mode (left); Rendering of the F-35B VTOL System (right)............ 1
Figure 2: Render of a Typical Low Cost Quadcopter..................................................................... 2
Figure 3: Shuttle Rocket Engine Gimbal Test ................................................................................. 4
Figure 4: Phase-Gate Project Framework Diagram ........................................................................ 7
Figure 5: Initial Concept Model of a Single GAVN .....................................................................11
Figure 6: Decision Matrix Used for Number of GAVN Units on Prototype Aircraft ............13
Figure 7: Conceptual Model for Three GAVN Aircraft ..............................................................14
Figure 8: Preliminary Control Algorithm for a GAVN System ..................................................18
Figure 9: Pseudocode Structure .......................................................................................................19

iv

Figure 10: GAVN COTS Layout Rendering .................................................................................21


Figure 11: Axial Pre-load Bearing Mounting Method 1st Stage ..................................................23
Figure 12: Cut-Away View of 2nd State Bearing System .............................................................23
Figure 13: Typical 4 Petal Futaba Servomotor Horn ....................................................................24
Figure 14: Comparison of Common Gear Types..........................................................................25
Figure 15: GAVN Servomotor Gear Assembly ............................................................................26
Figure 16: GAVN 50 Tooth Herringbone Gears ..........................................................................26
Figure 17: 50 Tooth 3D Printed Herringbone Gears ...................................................................26
Figure 18: Inner Gimbal Ring Interface .........................................................................................27
Figure 19: Inner Gimbal Ring FEA - Displacement Due to Max Engine Thrust....................29
Figure 20: Outer Gimbal Ring FEA - Von Mises Stresses Due to Max Engine Thrust .........30
Figure 21: Rendered Exploded View of GAVN Design Mark 1.5 .............................................31
Figure 22: Fully Assembled GAVN Test Stand ............................................................................31
Figure 23: GAVN Mark 2 Design Render......................................................................................32
Figure 25: Example of Y-Shaped Configuration ...........................................................................33
Figure 24: Example of T-Shaped Configuration ...........................................................................33
Figure 26: Preliminary Component Layout ....................................................................................33
Figure 27: Oversized Body Frame ...................................................................................................34
Figure 28: Design for GAVN Airframe ..........................................................................................34
Figure 29: FEA with GAVN Mounts Fixed ..................................................................................36
Figure 30: FEA with Airframe Center Fixed .................................................................................36
Figure 31: Power Distribution Wiring Schematic..........................................................................38
Figure 32: GAVN Prototype Signal Wiring Harness ....................................................................39
Figure 33: GAVN Component List.................................................................................................40
Figure 34: Ducted Fan Hanging with Weight Suspended ............................................................42
Figure 35: Ducted Fan Image Annotated with Measurements ...................................................42
Figure 36: Thrust-Voltage Data for Fan B .....................................................................................45
Figure 37: Mark 2 Test Stand Pattern .............................................................................................46
Figure 38: GAVN Test Stand Pattern .............................................................................................46
Figure 39: NI myRIO Connection Diagram; MXP-A & -B (left), MSP-C (right)....................47
Figure 40: Standard Servo Wire .......................................................................................................47
Figure 41: MotorTest VI Block Diagram (top) and Front Panel (bottom) ...............................49
v

Figure 42: FPGA Personality for PWM Outputs ..........................................................................50


Figure 43: KeyboardControl.vi Block Diagram .............................................................................51
Figure 44: KeyboardControl_v2.vi Block Diagram ......................................................................52
Figure 45: RT Main.vi Block Diagram ............................................................................................52
Figure 46: Prototype Aircraft Test Setup ........................................................................................54
Figure 47: Gantt Chart ......................................................................................................................64
Figure 48: GAVN Comparative Thrust Weight Energy Analysis...............................................65
Figure 49: Quadcopter Dynamics and Control LabVIEW Example VI ...................................66
Figure 50: Quadcopter Dynamics and Control Subsystem VI Integrations .............................67
Figure 51: EDF Model Rendering ...................................................................................................67
Figure 52: Servomotor Model Rendering .......................................................................................68
Figure 53: Servomotor Photograph .................................................................................................68
Figure 54: Inner Gimbal Ring FEA - Stress due to Max Thrust.................................................68
Figure 55: GAVN Mark 1 Design ...................................................................................................69

List of Tables
Table 1: GAVN Ball Bearing Design Requirements .....................................................................22
Table 2: Comparison of Purchased vs Required Bearing Values for 1st Stage .........................22
Table 3: GAVN Spur Gear Parameters ..........................................................................................25
Table 4: Relevant Average PLA Material Properties ....................................................................29
Table 5: GAVN COTS Components & Sources ..........................................................................41
Table 6: GAVN Team Expenditures ..............................................................................................59

vi

Introduction
Background
This project investigates the future of vertical takeoff and landing methods of actuation and
control. To understand where VTOL technology may be headed, it is necessary to first study the
history of VTOL, and then examine current VTOL technologies and their inherent problems. Finally,
the project looks to determine what technologies are available for the future that will serve as a solution
to these problems.
Vertical Takeoff and Landing
Vertical takeoff and landing (VTOL) is the ability of an aircraft to take off and land vertically
in a space only slightly larger than its own footprint, eliminating the need for a full-size runway. This
is a critical tactical ability for any aircraft, as it drastically decreases the cost in landing space required
and greatly increases its deployment flexibility. Most importantly, it allows the aircraft to hover in
place, providing numerous tactical advantages in both military and commercial applications. These
advantages could include targeting and firing directional weapons; operating cameras, radar, and other
line-of-sight imaging equipment like infrared and gamma ray systems from a fixed point; or precisely
lifting and placing a load for construction purposes.

Figure 1: F-35B in VTOL Mode (left); Rendering of the F-35B VTOL System (right)

Perhaps the most complex modern example of a VTOL aircraft is the Lockheed Martin F35B Lightning II Joint Strike Fighter. Though the F-35B (Figure 1) has a highly complex VTOL
system, it is only capable of vertical takeoff and landing purposes; it is not designed for continuous
hovering. As such, this expensive and complex system cannot take advantage of the most
advantageous feature of VTOL [2]. This example serves to illustrate that VTOL systems are not
1

universal, and in many cases are mission- and aircraft-specific. Furthermore, even in cases where the
system is of a high component complexity, the functionality may still be limited; that is, the complexity
is not necessarily linearly related to the functionality. This leads to the conclusion that just as a highly
complex and expensive VTOL system can provide limited functionality, it is possible for a simple and
cost-effective system to provide high functionality.
VTOL technologies are separated into two distinct methods of control: magnitude controlled
steering (MCS) and direction controlled steering (DCS), the latter being more commonly known as
thrust vectoring. Systems fall into one of these two categories based off of their approach to modifying
the thrust vector of the engine(s) either the magnitude or direction, hence the category names to
affect the orientation, position, velocity, and acceleration of the aircraft.
Magnitude Controlled Steering
MCS is almost exclusively employed on multi-rotor light unmanned aerial vehicles (UAVs),
primarily used in the consumer and light commercial sectors. The military and large commercial
operations have yet to make significant use of this technology, though these entities are showing
growing interest [3]. A classic example of MCS technology is a multi-rotor UAV called a quadcopter
(also known as a quadrotor, quadrocopter, and other similar variations; seen in Figure 2 below) A
quadcopter steers by changing the power to each fixed-orientation engine (rotor) independently: by
varying the magnitude of thrust at each of four points
on a square, the orientation of the aircraft changes.
This change in the aircrafts orientation results in a
corresponding change in direction of the overall thrust
vector, which allows the aircraft to move in any
direction. This particular method has the advantage of
not requiring any moving parts beyond the motors and
rotor blades to generate lift, reducing complexity and
weight, and thus cost and ease of manufacturing. This
is why a capable quadcopter can be built for under
Figure 2: Render of a Typical Low Cost Quadcopter

$500 dollars [4].

Unfortunately, the limitations of MCS technology limit its possible applications. Precise
control of the engines thrust is required for maneuvering; therefore, the engine must be capable of
rapid throttling control (RTC) such that the thrust magnitude from each engine changes quickly. This

requirement renders electric motors the only suitable current technology. Throttle response time (the
measurement of the time it takes an engine to respond to the pilots throttle set point command) in
an electric motor is only affected by the inertia of the stator and gearbox (if present) since signal travel
time is nearly negligible on the order of a few milliseconds. However, other types of engines such
as jet turbines and internal combustion engines have a much longer throttle response time: on the
order of 1-10 seconds peak time, and as long as 30 seconds settling time [5].
Electric motors are also usually powered via electrochemical storage mediums, which have a
far lower energy density than that of combustion fuels. For example, a typical lithium ion battery has
an energy density of 2 MJ/L as compared to jet-A fuel (kerosene) which has an energy density of 33
MJ/L [6]: jet fuel is 15 times more energy-dense than a typical lithium-ion battery. However, because
lithium is the 3rd lightest element and kerosene is a much heavier hydrocarbon molecule, a comparison
of energy density is not the critical factor to examine; particularly for VTOL aircraft, weight is the
most important factor. As such, a better comparison than energy density (energy per unit volume) is
the specific energy (energy per unit mass). The maximum specific energy of a lithium ion battery is
0.875 MJ/kg; the specific energy of jet fuel is 42.800 MJ/kg, nearly 50 times that of a lithium-ion
battery. Even when the energy conversion efficiencies of the systems that use these energy sources
are considered, a jet engine still has about 40 times the power per unit weight of a battery/electric
motor system. Thus, until battery technology makes a very significant leap forward, a jet fuel burning
jet turbine engine is still the clear choice when the power to weight ratio is the critical factor. However,
since jet engines are infeasible due to their long throttle response time, utilizing electric motors and
batteries is a necessary but major disadvantage of the MCS.
Another disadvantage of MCS is that the whole aircraft must tilt to a specific orientation in
order to move forward, forcing attached equipment such as cameras and other line of sight imaging
systems to rely on independent motion gimbals for stable imaging. These gimbals and leveling systems
add weight and complexity to an already underpowered aircraft. In addition, level flight is important
in situations where the aircraft is to be used for heavy lifting, meaning a vehicle that can maintain level
flight while countering dynamic winds is highly desirable factor in these situations.
Direction Controlled Steering
DCS was the first type of thrust vectoring to be invented and used and has shaped VTOL
vehicle design for decades through its numerous implementations. The origins of thrust vectoring can
be traced back to Dr. Robert Goddard and the early United States rocket program [7]. In his quest to

pioneer American rocketry, Goddard is attributed with inventing the liquid fuel rocket engine, the
multi-stage rocket, and the gimbaled liquid fuel rocket engine.
While Goddards newly invented liquid fuel rocket had major advantages over solid fuel
rockets, it did require some major innovation in the stabilization techniques used during flight. Solid
fuel rockets were designed to stabilize themselves by spinning rapidly, but this method would have
caused significant problems in the fuel flow and pump systems due to the centripetal forces generated
by the rotational velocity. As such, a new method of rocket stabilization that did not require spinning
was needed; Goddards solution was the gimbaled rocket nozzle. By pivoting the gimbal (and therefore
the attached rocket nozzle) in the opposite angle of the rockets trajectory misalignment, the rocket
could be kept on course. Later on in the program, when multiple liquid rocket engines were used on
a single stage (as in the Redstone and Titan missiles), the
gimbaled engines added another important feature: if one
or more engines in a multi-engine system lost some or all
of their thrust output, the others could compensate by
changing their gimbal angle to offset the loss, thereby
keeping the thrust vector pointed in the correct direction.
This new feature made the gimbaled liquid rocket engine
capable of not only increased stability, but also added
redundancy and safety in the event of failure. The gimbal
is seen today as a crucial technology in the space program.
Pictured at left (Figure 3, [1]) is one of the Space Shuttles
main engines mounted independently on a test bed, during
Figure 3: Shuttle Rocket Engine Gimbal Test

a full thrust burn at a maximum gimbal angle of 20 from

center. The gimbal mechanism for rocket engines (which usually takes the form of a universal joint
actuated by hydraulic cylinders) proved that control and stabilization of a VTOL aircraft can be
performed via DCS.
In early VTOL aircraft, the exhaust nozzles of jet turbine engines were made to pivot through
moving ductwork (though the mechanisms construction took a much different form than a gimbaled
rocket engine), thereby re-directing jet exhaust downward. The first successful VTOL jet turbine
aircraft was the McDonnell Douglas AV-8B Harrier II fighter, commonly referred to as the Harrier
Jump Jet [8]. The Harrier series of aircraft were the first time ducted thrust vectoring had been used
successfully in a jet aircraft, and though the end result was a capable aircraft and a milestone
4

achievement in aviation technology, the Harrier program also showed that VTOL aircraft are
inherently difficult to control manually. As discovered through years of testing and a few tragic crashes,
pilots had a great deal of difficulty learning to control the first-generation Harrier in hover mode. It
had no automatic stabilization, instead forcing pilots to completely control it manually; one pilot
claimed that flying the Harrier in hover was like trying to balance an elephant on an exercise ball with
a remote joystick [8]. In an attempt to fix the issue, the control sticks range of motion was limited,
and a delay was added between the pilot input and the system response to make the system more
sluggish and prevent human overcorrection. Later generations of the Harrier incorporated an
automatic stability assistance system to further fix the problem, but it was disabled after receiving
negative reviews during pilot testing. As a result, the pilots were simply told to try harder by their
superiors, and so the Harrier pilots became known as some of the best pilots on the planet.
The next generation of VTOL fighter the aforementioned F-35B came a number of years
after the Harrier. The F-35B improves upon the Harrier and its stability issues by combining a variable
geometry nozzle with a ducted fan, improving stability in VTOL mode by eliminating the difficult to
control rotating side ducts. Additionally, the VTOL mode of the F-35B is fully automated, so much
so that the pilot only has to pull a single lever. The variable geometry jet exhaust nozzle is a form of
DCS, and is similar in function to the gimbaled rocket engines used on spacecraft. A good example
of this is the Lockheed Martin F-22 Raptor and the Eurofighter Typhoon II. The variable geometry
nozzle has moving panels that are individually actuated to change the shape of the nozzle and in turn
redirect the exhaust jet. Some aircraft can only redirect the exhaust along one axis (generally the pitch
axis), but the F-35B can actuate its rear nozzle in two directions to provide hover stability.

Stakeholders and Needs


Potential stakeholders of a technology that improves VTOL capability would be companies
already producing aircraft with similar technology such as Lockheed Martin, Boeing, Raytheon,
Northrop Grumman, BAE System, NASA, SpaceX, Orbital Sciences, the Department of Defense,
Department of Energy, and the US Military. The technology is applicable to the military and
commercial space and aerospace industries, the power, oil and gas industries, construction, and search
and rescue industries.
The primary needs of the project are to:

Demonstrate the viability of constant-magnitude thrust vectoring

Maintain level flight during rotational and translational maneuvers


5

Maintain an absolute position against dynamic external forces

Provide a flexible platform for various existing and future engine types

Match the maneuvering capability of comparable VTOL technologies

Secondary needs are to:

Land safely after a catastrophic engine failure

Exceed current technology in


o Maneuver capability
o Maneuver consistency
o Maneuver execution speed

Problem Statement
There is a need for a new method of aircraft steering that is compatible with both existing and
future types of engines that are not usable with current VTOL methods. To facilitate the expansion
of VTOL technology, a new method of vectoring the thrust of an engine nacelle is required.

Methods
Project Framework
In order to provide a more rigid and defined structure for the design process, the phase-gate
(also known as stage-gate) development framework was implemented [9]. This consists of five main
development phases with five corresponding checkpoint gates (hence phase-gate) at which the
development team must show evidence of having met certain requirements before they can proceed
onto the next phase. While exact titles vary from organization to organization, the main objectives for
each remain the same. For this project, the five phases are as follows:
1. Scoping & Feasibility
2. Architecture & Planning
3. Design & Development
4. Testing & Optimization
5. Launch

Figure 4: Phase-Gate Project Framework Diagram

A zero phase, entitled Discovery & Investigation, is often also included, as seen in the above
chart [10]: it is the shortest phase of all, as it is limited to concept generation and general
communication with product users and component suppliers. Concept generation is limited to
product ideas and does not include creating drawings or diagrams of the idea beyond what is necessary
to convey the idea to managers. Following the zero phase is the first gate: this is where the project
management team decides whether or not the development should proceed. The equivalent senior
design actions for this phase are the formation of teams, the selection of a concept, and the location
of a professor to mentor the team.
If the concept is considered viable it will pass through the first gate and on to the first phase,
Scoping & Feasibility. This is when brainstorming sessions specific to the concept are held, with the
goal of establishing a framework for the future design, including basic analysis of its fundamental
attributes and of its market potential. It is highly advisable to conduct research into comparable
technologies (including directly and indirectly competing designs) as this will help shape both the
design goals and provide an estimate of the threat level presented by competitors. Once the above
work has been completed the project moves on to the second gate, where management again analyzes
the viability of the concept and determines if it should move forward into the much more intensive
second phase. The research and basic analysis (including budget estimates) required for the senior
design preproposal constitute the work relevant to this phase, while submitting the preproposal and
receiving approval to proceed satisfies the requirements of this gate.
The main objective of the second phase, Architecture and Planning, is to fully detail the
guidelines for the concept and its development process. This stage includes but is not limited to
expanding the product definition into specific needs and specifications, finalizing the business case,
7

and outlining the timeline for the rest of the development process. Most work conducted during fall
term covered phase two activities, including research into both current and historical gimbal and
VTOL technologies as discussed above, outlining the requirements for the GAVN as well as the
demonstration prototype aircraft, modeling a GAVN concept for demonstrative purposes, and
creating a Gantt chart to schedule development over the course of the project. The fall proposal
presentation and this report are the materials submitted for evaluation at the third gate.
In the Design and Development phase (number three), the plans created in previous steps are
executed. The overarching goals set in phase two are broken down into SMART (Specific, Measurable,
Attainable, Relevant, and Timely) milestones so as to provide clear indicators of progress during the
development process. This phase can also include alpha testing of each system as it comes online.
Over the course of winter term, work will proceed on all facets of the project, including detailed
component modeling, dynamics assessments of the GAVN units and the prototype aircraft, airframe
statics analysis, and control system and user interface programming. The requirements for approval
to proceed at gate four are fully realized component designs (including supporting documentation)
and the physical prototypes necessary for phase four, Testing and Optimization.
Phase four covers all aspects of testing, optimization, integration, and any other related efforts
leading to the gate five deliverable of a ready-for-market product. These activities are comprised
primarily of lab-based internal testing (including any necessary design changes) and field testing in
environments representative of the expected use case. Field testing can also include consultations with
possible customers in order to analyze their interest level in the product, their preferred product
attributes, and the likelihood of them purchasing the product. Manufacturing processes are also
outlined during this phase. Limited market testing is sometimes conducted during this phase to assess
and update sales forecasts and test proposed marketing plans.
With the finished product ready for launch, phase five consists of finalizing the marketing
strategy, setting initial production volume, and training manufacturing, sales, and support staff on the
new product. As it pertains to senior design, the GAVN design projects goal is to demonstrate the
viability of the design, not to launch a final, ready-for-market product, and so work in phase five will
be minimal. A prototype aircraft with full flight capabilities was the intended deliverable for the final
senior design presentation.

Requisite Skills and Modern Engineering Tools


This project demanded competency in multiple varied disciplines of mechanical engineering
in order to achieve full success. Dynamics and statics were critical, as the torques generated by
articulating the GAVNs result in significant stress on the airframe and motion through threedimensional space. The utilization of material from mechanics of materials, aircraft dynamics and
performance, and advanced dynamics was required. Creating a structurally sound fuselage was
necessary in order to protect the electronics mounted to it; the goal was to have these electronics run
multiple algorithms in order to sense both the aircraft and the individual GAVNs positions and
velocities, calculate the necessary adjustments to respond to movement commands and external
disturbances, and modulate the response to maintain stability.
Software tools greatly assisted in this process. Computer-aided design tools facilitated rapid
modeling (including digital analysis) and additive manufacturing, which was ideal for the small,
intricate parts that the prototype incorporated. Finite element analysis was utilized to model stresses
in the airframe and GAVNs in order to detect possible failure points. Finally, advanced
microcontrollers and sensors were installed to provide the calculation rates required to analyze aircraft
dynamics in real time and thereby achieve stable flight, while graphical programming software
simplified both control system simulation and final coding and testing.

Concept Generation
Specifications
The specifications for this project were formed from the needs as stated above. Each
specification has its own weight on the development of the concept; this weight was proportional to
that specifications overall importance. The separate needs of maintaining level flight during yaw and
horizontal translation and maintaining level flight during dynamic external forces can be combined
into one specification: the concept should be able to maneuver while keeping the aircraft body level
within a range of 5 (which current systems like quadcopters are incapable of). In order to hold its
absolute position, the craft should be capable of knowing its position in space within one to three
inches so that it can detect when it has left the desired position if in hover mode. One of the highest
weighted specifications of the concept is that it is capable of utilizing multiple propulsion systems;
therefore, the concept must be controlled through a system that maintains functionality independent

of propulsion source and presents no obstructions along the thrust vector when in the nominal hover
configuration.
The secondary needs of the system include the ability to maintain stability in the event of an
engine failure; therefore, the system must have some redundancy to allow for stable (if limited) flight
and landing. The remaining specifications come from comparisons to current generation quadcopters,
which are very capable in terms of maneuverability and rapid movement. The system developed by
this project must meet or exceed the capabilities of quadcopters in all aspects. This includes but is not
limited to: performing maneuvers with greater speed, reliability, and repeatability; performing
maneuvers that compete with the capabilities of current systems, including flips, rolls, etc.; rapidly
reversing direction in less distance than competing systems; and improving the flight duration and
range of comparable quadcopters.
The final specifications that affect the development of this concept are those regarding existing
codes and standards related to the design and aviation fields, including ASME, NASM, ASM, ANSI,
and FAA regulations. These include specifications about bearings, gears, materials, and testing
requirements, among other things. One immediately relevant code is the FAAs AC 91-57 Model
Aircraft Operating Standards, which provides conditions for flight testing of the system, including
maintaining a flight ceiling of 400 feet and not flying within range of an airport or other location
containing full size aircraft. Other codes were also cited for the design and selection of the GAVNs
gears and bearings, as well as the prototype aircrafts airframe material the specific codes and the
sections cited will be discussed in later sections.

Concepts
After considering the needs and specifications, the team created the concept of the Gimbal
Actuated VTOL Nacelle (GAVN). The concept utilizes the theory that an aircraft can be controlled
through direction-controlled thrust vectoring. The system consists of a bi-axial gimbal into which any
compatible propulsion system may installed. The design of the gimbal is such that the method of
propulsion is not stipulated by the system (beyond physical compatibility requirements), and control
is achieved solely through changing the direction of the thrust by actuating the gimbal. This design
facilitates a large degree of flexibility in the propulsion system: it can utilize a brushless electric motor,
which is capable of changing its thrust rapidly; a jet turbine, which can only change its thrust output
slowly; or futuristic technologies such as ion engines. Ensuing design decisions are based off this idea

10

of a control system functions by maintaining constant thrust magnitude and instead varying the thrust
direction.

Figure 5: Initial Concept Model of a Single GAVN

Within this overall concept were multiple design elements that required evaluation to
determine the ideal choices based off of the project needs and specifications. Five different initial
design elements were initially considered.
The first and second elements were the number of GAVN units per aircraft and the shape of
the craft, and are closely related in that the shape of the craft is significantly dependent on the number
of units being used. Four different iterations of these elements were evaluated: crafts utilizing one,
two, three, and four or more GAVN units.

One GAVN: this craft has the unit placed at the center of the fuselage. This system is
the simplest of all the options due to the propulsion being at the center of gravity, and
there being only one unit to control; however, the system would be limited in its
maneuvering capabilities due to only one point of control.

Two GAVNs: this craft would have more capabilities than the single unit concept, but
the complexity is also increased. In this concept the units would be placed along one
axis through the center of gravity.

11

Three GAVNs: this design provides additional control over both the one and two unit
systems, again with increasing complexity. The largest advantage of a system that uses
three units is the existence of redundancy that is not present on the other systems. The
three unit system would likely place a unit on either each face or corner of a three sided
aircraft.

Four or more GAVNs: This system is similar to the three unit system in terms of
maneuverability and stability while providing an increase in redundancy; however, the
complexity is greatly increased.

After deciding how many GAVNs would be used on the proof of concept aircraft, the next
design element chosen was the propulsion system. Limiting the options to readily available current
technologies, the choice was between electric motor driven propellers, ducted fans, and jet turbine
engines. Jet turbines provided more thrust than any other options; however, they were more
expensive, complicated, and dangerous. The propeller and ducted fan options were similar in that they
both use electric motors, with the greatest differentiation coming from the fact that ducted fans are
more efficient in controlling the exact direction of the thrust output from the system. Electric motor
driven systems are also the standard in small-scale aircraft such as quadcopters, whereas jet turbines
are rarely, if ever, used in these applications.
The final two design elements considered were the range of motion for the gimbal servos and
the payload capacity of the aircraft. The range of motion of the gimbal consists of the angle range of
each axis individually. Each axis has a potential range up to a full 360, but needed to be limited so
that the thrust vector does not point into the aircraft body. Beyond that requirement, the primary
objective was to design a system with the most capabilities for the lowest degree of complexity: as
such, the decision was made to use a highly capable servo with a simple mechanical link to the gimbal
rings, thereby avoiding the use of any complex systems that would provide angular actuation
amplification between the servo and gimbal.
Finally, the payload capacity of the aircraft as a design element is affected by the engine choice
for the GAVN units as well as the weight allowance needed for the power source and other
components. This design element was codependent on the propulsion selection in that the larger the
propulsion system used is, the larger the power source must be; therefore, optimization is required to
decide on the most efficient sizes of each and the resulting payload capacity. As such, a specific payload

12

capacity requirement for this prototype was not initially set; instead, the goal was to minimize aircraft
weight (maximizing thrust to weight ratio) wherever possible.

Concept Evaluation
At the conclusion of the fall term of senior design, three of the five design elements had been
evaluated for used on the proof of concept aircraft. The first decision made was regarding the number
of GAVN units to be used on the prototype, utilizing criteria such as cost, maneuverability,
redundancy, stability, and simplicity. The one unit system was ranked with threes for all criteria to
serve as a baseline. Each other system was then compared, criterion by criterion, to the one unit
system: if the system was judged to perform better than baseline, it was assigned a four or five
(depending on how much better); if the system was worse, it received either a one or a two. Each
criterion is given a weight based on the importance based on the needs and specifications of the
project. The values assigned to each criterion were then multiplied by their respective weights, after
which a summation of all criterions for each design was taken. The design with the highest total score
represents the ideal design element.

Figure 6: Decision Matrix Used for Number of GAVN Units on Prototype Aircraft

As shown in the decision matrix in Figure 6, the three GAVN system was determined to be
the best option for the proof of concept aircraft. No single iteration had the best option with regard
to all criteria, but the advantages of the three unit system outweighed the disadvantages. A triangular
shape was also chosen for the aircraft hull as a result of this decision.

13

Figure 7: Conceptual Model for Three GAVN Aircraft

An evaluation was also carried out to determine which propulsion source should be used.
Choices were evaluated based on availability, capability, complexity, and cost. This analysis showed
that the cost and complexity of a jet turbine engine was too great for it to be used on the prototype
aircraft, reducing the options to either an open propeller or ducted fan system. Both of these systems
had similar availability, complexity, and cost. The decision was made to use a ducted fan system
because of its ability to efficiently direct thrust.
In the winter term, the final two design elements (the range of motion for the gimbal servos
and the payload capacity of the aircraft) were determined. As mentioned previously, each axis has a
theoretical range of motion of a full 360; this full range was determined to be unnecessary for the
fulfillment of the project objective. Instead, it was decided that a range of 180 per servo was a more
realistic minimum for this prototype aircraft. This decision was made for two main reasons: first, 180
is the most common range for commercial off the shelf (COTS) servos, which simplifies component
selections. Second, basic analysis of the flight patterns intended for this project indicated that the need
for a single servo to rotate more than 90 from nominal was virtually nonexistent: in almost all cases
where a GAVN module would rotate more than 90, a different orientation of the aircrafts GAVNs
could be used and would achieve the same result without an over-rotation.

14

The payload capacity of the aircraft was the last design criteria finalized since it is an element
that is directly tied to choice of engine. It was determined that for prototyping purposes that there
would be no distinct payload in this aircraft: instead, the payload will be set as the weight of the
battery, myRIO, electronic speed controllers (ESC), sensors, wiring, and the aircraft frame. With all
the major design elements decided, development then proceeded to detailed design and component
decisions required for the creation of this aircraft.

Design Process
Theoretical Modeling
Dynamics
The forces and moments acting on the vehicle are observed in terms of coordinate reference
frame transformations. Each GAVN unit has its own reference frame with the origin at its center of
gravity; each GAVNs reference frame is parallel to the vehicles reference frame. This approach
simplifies the representation of the GAVNs thrust as constant force acting along the GAVN-frame
z-axis, shown in Equation 1 (Note: positive z-direction is downward).
= [0 0 ]

(1)

The GAVN unit has two degrees of freedom: rotation about its y-axis (pitch) and rotation
about its x-axis (roll). The GAVN thrust vector can be transformed back into its rest state using
rotation matrices, shown in Equations 2 (pitch), 3 (roll), and 4 (yaw). Multiplying the vehicle reference
frame by both rotation matrices will transform a vector to be parallel to the GAVN reference frame;
transforming in the opposite direction (GAVN to vehicle) is performed by multiplying the vector by
the inverse of the rotation matrices, as seen in Equation 5.
C
{} = [ 0

1
0 S
1
0 ], {} = [0
0 C
0

0
C
S

0
C
S ], {} = [S

0
1

= 1 = ( )

S
C
0

0
0]
1

(2),(3),(4)
(5)

The same method is used when calculating the moment of inertia of each GAVN. The
moment of inertia is the volume integral of the mass density of the GAVN, shown in Equation 6.

15


= [

2 + 2
] = () [

+ 2

]
2
+ 2

(6)

Since the GAVNs reference frame origin is located at its center of gravity, planes of symmetry
exist which eliminate all products of inertia, resulting in Equation 7:

G = [ 0
0

0
0]

(7)

The GAVN moment of inertia is then rotated to be parallel to the vehicle reference frame
using the product of both rotation matrices, R:
=

(8)

Finally, parallel axis theorem is used to translate the GAVN moment of inertia to the center
of gravity of the vehicle (Equation 9). Since the center of gravity of both the GAVN and the vehicle
are coplanar, the displacement vector between them lacks a value corresponding to the z-direction;
the GAVN inertia is translated on the x-y plane.

2
= + [( )3 ] = + [
0

2
0

0
0 ]
2
+ 2

(9)

The total moment of inertia of the vehicle is the sum of all moments of inertia about its center
of gravity, which consists of each GAVN and the vehicle body itself. The moment of inertia of the
vehicle body has not been analyzed to date, as its final shape and dimensions have not been finalized
at this stage of the design process.
The equations of motion of the vehicle are derived by using Newton-Euler equations, shown
in Equations 10 and 11. These equations describe the behavior of the vehicle in response to external
forces (F) and the moments (M) caused by those forces. The moment equation (Eqn. 11) also includes
contributions from the spinning fans in the form of their angular momentum, HR. This rotor angular
momentum is already rotated to the body frame and translated to the vehicles center of gravity as
previously described. If the rotor contribution to the moment equation is negligible, it can be omitted.
This will be determined once the vehicle is built its performance is evaluated.

= + ( ) = [ ] + [ ] [ ] = [ ] + [ ]

= + ( ) = + + ( ) + ( + )

(10)
(11)

16

It should be noted that the moment of inertia is time-dependent due to the changing angles of each
GAVN, which directly influence the total moment of inertia. This adds a layer of complexity to the
time-derivative of angular momentum, seen in Equation 11. Rearranging these two equations provides
the rate of change of both translational velocity and angular velocity, shown in Equations 12 and 13,
respectively.

]
[ ] = + [

(12)

(13)
= [ ] = 1 [ ( ) ( + )]

The rotational kinematic equations of a vehicle with six degrees of freedom are shown in
Equation 14. These relations are derived from transformations between the vehicle reference frame
and the earth reference frame described in Stevens (2003).

1 sin tan

cos
[ ] = [0
0 sin sec

+ ( sin + cos ) tan


cos tan
sin ] [ ] = [
cos sin
]
cos sec
( sin + cos ) sec

(14)

All of this dynamic modeling of the aircraft will be used to control its flight and stability. A
preliminary layout for this control scheme is shown in Figure 8. The general control process for the
GAVN system is similar to other VTOL vehicles, especially for attitude stability. The primary
difference is how the vehicle physically achieves stability/flight: thrust vectoring of the GAVNs.

17

Figure 8: Preliminary Control Algorithm for a GAVN System

The aircrafts current state (lateral and angular positions and accelerations) were to be
measured by various sensors, such as a gyroscope, accelerometer, rangefinder, and optical flow sensor.
These measurements were to be used in the microcontroller to calculate the necessary forces that are
needed to move the aircraft to a desired state. The microcontroller would constantly record the angular
positions of each GAVN, while outputting to them the adjustments needed to produce the forces to
reach the aircrafts desired state. The output of the microcontroller would control the pitch and roll
servos and the propeller motor of each GAVN. Shaft encoders in each GAVN will provide
measurements of each motors speed to ensure that they all maintain parity. The adjustments to the
GAVNs would result in a net force causing the aircraft to move, which will be measured by the inertial
positioning sensors, restarting the cycle.

18

Code
The tricopter utilized a myRIO microcontroller provided by National Instruments (NI) and
programmed using NIs graphical programming language, LabVIEW. Built into LabVIEW are various
example programs called virtual instruments (VI), including one that simulates the flight of a
traditional X-shaped quadcopter (shown in Appendix A, Figure 49). This quadcopter example also
implements PID loops to control the vehicles roll, pitch, yaw, and altitude. This example VI was
planned to be implemented and adapted to the tricopter GAVN design, which was possible due to its
use of LabVIEWs MathScript RT Module (Appendix A, Figure 50). The MathScript node allows the
use of MATLABs m-file syntax in a LabVIEW VI. The MathScript node in this simulation contained
the flight dynamics of the vehicle, allowing it to be easily modified to whatever the user desires. This
node was planned be modified to describe the tricopter according to its dimensions and the equations
of motion shown in Eqns. 12 and 13. The preliminary MATLAB script that would have been used in
the MathScript node is shown in Appendix C: Tricopter Flight Dynamics Code.
Other adjustments needed to be made to the code to improve its organization and to keep
with the modular spirit of the GAVN system. In this pursuit, the subsystem model will be split into
two main components: a six degree-of-freedom flight dynamics model, and an aircraft components
actuation system. The relationship between these components is outlined in Figure 9. The flight
dynamics model contains the rearranged Newton-Euler equations described by Eqns. 12 and 13. Its
inputs are the resultant forces and moments acting on the vehicle, the vehicles moment of inertia,
and the angular momentum of the rotors. The output from the flight dynamics model is the six state
derivatives corresponding to translational and rotational acceleration in the body frame. The
components actuation system takes the information from the flight dynamics model and translates it
into proper servo angles that cause the three GAVN units to produce the necessary resultant forces
and moments. It will also compute the correct PWM signal for each servo.
Flight Dynamics
Model

Aircraft
Components
Actuation

Servo PWM
Forces/Moments, Inertia,
Rotor momentum

Figure 9: Pseudocode Structure

19

Detailed Design
In order to ease optimization and component selection for the prototype, a flexible template
was created in Microsoft Excel (Figure 48 in Appendix A) that was used to quickly estimate the
performance metrics for any given set of components and their respective specifications in any
configuration. This ensured that the components selected would result in an aircraft that was best able
to fulfill the requirements set in the project specifications.
To obtain the flight duration estimates, battery specifications such as weight, maximum
constant current, peak current, nominal operating voltage, and capacity were used to obtain derivative
values, including maximum constant power output and energy density. The template then calculates
total parameters for a battery cluster (as specified by the total number of batteries and the number of
series they are configured in).
To estimate power usage, thruster specifications like maximum thrust, power draw at max
thrust, nominal operating voltage, and weight were input, then multiplied by the number of total
thrusters on the aircraft. A goal power to weight ratio was obtained from research into remote-control
hobbyist helicopters, and used in conjunction with the previous data to calculate the target total weight
of the vehicle. The previously calculated battery and thruster weights were subtracted from this goal
to estimate the maximum allowable weight of the airframe and potential payload.
Finally, the maximum flight duration at max thrust and in hover mode were computed:
maximum thrust flight time was calculated directly from the battery and motor specifications, while
hover mode time was obtained by approximating the motors power usage for different thrust outputs
as a linear curve (with a safety factor to account for nonlinearity) and solving for the point where the
thrust output was equivalent to the vehicle weight.
GAVN
Initial Brainstorming and Configuration Design
The initial design of the GAVN unit originated with concept models and sketches generated
in the fall term as the foundation (Figure 7). In winter term, it was adapted around the required COTS
hardware components of the prototype GAVN, including the engine, the pitching axis (inner gimbal
ring) servomotor, and the roll axis (outer gimbal ring) servomotor. Utilizing both published
dimensions and manual measurements, the COTS components were modeled in Solid Edge ST5,
added to a prototype assembly, aligned with a datum plane, and spaced according to an initial estimate
of component sizes (Figure 10). The GAVNs custom components were then modeled around the
20

digital COTS part models and assembly layout to ensure accurate sizing. After being properly
constrained, the model was also rotated and actuated along its motion axes within the assembly in
order to ensure the fan cowling would clear the gimbal rings, and to model clearances for power and
data cables.
Initial designs had the servomotors directly
attached to their respective shafts in order to save space
and weight, as well as reduce the required number of
parts. However, since these servomotors internal gear
trains are rated to handle the maximum torque of their
motor and not significant return torque, and that in the
event of a crash a GAVN gimbal ring may experience a
sudden rotation event generating an input torque far
exceeding the servomotors capabilities, this direct-drive
configuration

was

removed.

Instead,

simple

Figure 10: GAVN COTS Layout Rendering

mechanical linkage was instituted to help prevent servomotor damage, as gears, pulleys, and similar
mechanical linkages can be more easily replaced than the motors. With the GAVNs frame designed,
the next step was to select the remaining COTS components required for operation, including ball
bearings, gears, and mechanical fasteners.
Bearing Selection
It was important to include some kind of bearing to facilitate smooth rotational operation at
the inner-outer ring and outer ring-airframe mount interfaces. Selecting the proper bearing type out
of the multitude available was simple once a basic list of performance requirements was established.
As a general engineering rule of thumb in aircraft design, it is ideal to keep weight at a
minimum while maintaining a reasonable factor of safety and meeting other requirements. Given the
maximum thrust load of 8.33N (determined from the engines maximum achieved test thrust test
results can be found later in Thrust and Voltage Testing), the maximum continuous radial load any
one bearing will experience is half of the max thrust (two bearings per gimbal axis), or 4.17N.
Implementing a factor of safety of 2.0, the maximum radial continuous load is roughly 9N. A friction
journal bearing was initially considered, especially because of its simple design and low cost; however,
the shafts material, PLA (polylactic acid), is a high friction material, and would not function well in a
friction bearing as the shaft material. Implementing a metal shaft in the gimbal rings would have

21

facilitated the use of a friction bearing (due to the metals significantly greater hardness, compared to
PLA), but also would have created additional design complexity and added significant weight.
As such, the decision was made to utilize ball bearings, which are more durable and reliable
than friction bearings. McMaster Carr was identified as the most convenient supplier for bearings, and
their selection system made narrowing down and finding the appropriate bearings extremely
straightforward. The table below (Table 1) includes the other functional requirements for the GAVNs
bearings.
Specification
Radial load
Axial load
Rotation speed
Life

Value
9.00
2.00
300.00
1E9

Unit
N
N
Rpm
# of rotations

Table 1: GAVN Ball Bearing Design Requirements

The gimbal design inherently requires each rotational stages shafts to support the entire
weight of that stage and any further stages. Due to the differences in mass (and thus, inertia) of the
inner and outer gimbal rotating components, the outer rings shafts and bearings have much higher
load requirements than those of the inner gimbal ring, and need appropriately larger diameter shafts
and bearings. Since inner diameter plays a crucial role in bearing selection, this was the last piece of
information required to make a selection. The outer diameter of each shaft was determined by material
stress analysis, after which the outer diameter was reduced to be as small as possible to reduce size
and weight. Axial load, radial load, and rotation speed constituted the remainder of the functional
requirements. The bearing lifespan was not listed on McMaster Carrs product details list; in its place,
a tolerance rating of ABEC-5 was listed (the ABEC-5 code specifies a design life of 1E5 multiplied by
the maximum rotation speed). A table showing the required versus actual ratings of the inner gimbal
ring (1st stage) bearings is below (Table 2). It can be seen that the selected bearing vastly exceeds the
requirements in every way. A similar specification and selection process was used for the larger 2nd
stage bearings.
Specification
Radial load
Axial load
Rotation speed
Life

Required Value
9.00
2.00
300.00
1E8

Actual Value
302.50
130.20
63,000.00
1E10

Unit
N
N
Rpm
#of rotations

Table 2: Comparison of Purchased vs Required Bearing Values for 1st Stage

22

The bearings could be mounted in many different ways; however, the axial pre-loaded bearing
mounting method provides the best configuration for ease of manufacturing, ease of replacement,
and reliability. This method requires creating
a step in the shaft that presses axially along
the inner race of the bearing, while the outer
race is supported by the bearing block on the
other side (Figure 11). Often, machines make
use of at least two bearings on a single shaft,
and the GAVN design incorporates this
symmetric redundancy.

Figure 11: Axial Pre-load Bearing Mounting Method 1st Stage

In order to keep the bearings from falling out during operation and to keep the shaft from
spinning within the bearings, axial pre-load was designed into both stages of the gimbal. The normal
force required for this pre-load was minimal due to the high coefficient of friction of PLA. It was
calculated to be 1.1N, but was rounded up to 2.0N for a factor of safety. The inner gimbal stages
preload force comes from the elasticity of the PLA and the design of the outer gimbal ring, which was
modeled to have a slight (3 from centerline) inward camber. This causes the U-shaped part to deform
elastically outward like a spring as the inner gimbal ring (with bearings already fitted onto the shafts)
is pressed into place. As seen in Figure 11, the compressive force pushes from the outer stage ring
onto the two opposing outer bearing races on either shaft of the inner ring, which in turn presses on
the shaft steps. In the outer gimbal ring bearing stack, the shaft has a threaded hole in its end which
allows for a bolt screwed in from the back of the shaft to provide the correct axial load (Figure 12).
The bearing axial pre-load can be changed by adjusting the tensioning bolt.
When the prototype was assembled the
bearings worked impressively well. As a
rudimentary test, the inner gimbal ring was
attached (through the bearings) to the outer
gimbal ring and a small impulse was delivered
to one side of the inner gimbal ring. This was
enough to cause the ring to rotate many times
before stopping. While the bearings used may
be over-designed for this prototype, this level
Figure 12: Cut-Away View of 2nd State Bearing System

23

of quality would be necessary for any mass-produced GAVNs.


Gear Selection and Design
In the initial iteration of the design, typical RC hobby servomotor bar linkages were utilized:
these linkages provide a reliable transfer of power, an easy method of connection and are easily
shortened or lengthened to change rotation ratios. Typical RC hobby servomotors are sold with a 2,
3, or 4 petal horn (Figure 13) which is
simply an internally splined shaft with arms
extending radially outward, each having
several holes along each arm for mounting
the linkages. Typically these linkages (push
rods) are fixed to a working surface.
However, since the servomotors selected
could provide ample torque at a 1:1 transfer
Figure 13: Typical 4 Petal Futaba Servomotor Horn

ratio, and after modeling the servo horn

linkage it became apparent that there was no way to use push-rod linkages without severely restricting
the range of motion (given the radius of the servo horns available), the decision was made to use a
pulley system or gear system to transfer power instead. Pulley systems were quickly thrown out simply
because of they have a tendency to slip and require free space between parts, going against the goal of
keeping the GAVN design as space and weight efficient as possible. Pulleys also had other
complications, such as the need for tensioning and the high wear rate of belt materials in conditions
where the direction of rotation reverses. Based on all of these factors, gears were selected as the ideal
solution.
An online gear calculator found on Stock Drive Internationals website was a useful tool to
obtain preliminary calculations of the correct gear parameters. Spur gears were initially selected
because of their effectiveness and simplicity: the driving parameters of center to center distance, gear
ratio, pressure angle, face width, and hub diameter were input to the calculator, which then provided
the diametral pitch, number of teeth, and pitch diameter. The results are detailed below in Table 3.

24

Parameter
Center to Center Distance
Pressure Angle
Gear Ratio
Face Width
Bore Diameter
Diametral Pitch
Number of Teeth
Pitch Diameter

Small Gear
16 mm
20
1:1
5 mm
5 mm
0.5 teeth/cm
32
16 mm

Large Gear
25 mm
20
1:1
5 mm
10 mm
0.5 teeth/cm
50
25 mm

Table 3: GAVN Spur Gear Parameters

The original plan was to purchase precision machined plastic gears from Stock Drive
International, and permanently adhere them to the PLA shafts with cyanoacrylate adhesive. However,
the high cost and unfavorable design of these gears was deemed unacceptable. Instead of downloading
involute geometry 3D CAD models of the gears (as provided on Stock Drives website) to modify for
compatibility and 3D printing in
the

MEM

Departments

Undergraduate Lab printers, it


was decided (after receiving
advice from the lab manager,
Brandon

Terranova)

that

herringbone gears would be a


better choice than spur gears.

Figure 14: Comparison of Common Gear Types

To quickly summarize: a herringbone gear is a double helical gear with no gap between the
opposing rows (as seen in Figure 14) that offers much higher contact area over a much larger range
of rotation than standard spur gears; the design of a spur gear is such that at any given time, just a few
teeth or only a small part of a tooth are actually in contact between the two gears. Herringbone gears
are also nearly impossible to slide apart, and the larger contact area of each tooth drastically reduces
the mechanical stresses on each tooth.
Creating a 3D model of a herringbone gear turned out to be particularly challenging, as it
involves a swept geometry about a curve which is tangent to the gears axis of rotation yet parabolic
in plane with the tooth face width. This challenge was overcome by using offset planes and projecting
lines onto the surface of a cylinder, and then using that projected parabolic line as the sweep command
rail profile, with the spur gear acting as the sweep profile. This procedure generated a helical gear and
was simply mirrored to produce the herringbone gears (Figure 16).

25

Figure 16: GAVN 50 Tooth Herringbone Gears

Figure 15: GAVN Servomotor Gear Assembly

There still remained the issue of how to connect the other gear to the servomotor. Since the
motors were expensive and had a long lead time, permanently affixing the gears onto the servomotors
with adhesive was not an option. Instead, the horns included with the servomotors were sacrificed to
use their bolt-compressed splined hub attachment system, which was molded to interface with the
spline on the servomotor itself. The horns were modeled using caliper measurements, and the gear
designs were modified to include a cutout that would allow a modified servo horn to be press-fit into
the gear. This design only required that the gears and modified horns be glued together, both of which
could be easily replaced if necessary. The assembly (Figure 15) shows the modified horn, the gear with
the special cutout for the horn, and the servomotor.
The 50 tooth gears printed successfully (Figure 17). The teeth
were adequately spaced, well-defined and solid, and required little to
no additional hand-tooling. The first batch of smaller 32 tooth gears,
however, had slightly deformed teeth with excess PLA filament
between them. This was intially thought to be a complication of the
small geometries and the printers 100 micron resolution: the channel

Figure 17: 50 Tooth 3D Printed

between the teeth was small enough that even small amounts of excess

Herringbone Gears

PLA would block the channel entirely. When installed in the GAVN assembly (Figure 18), the inner
gimbal ring would rotate freely due to the gear teeth failing to mesh together.
These deformations proved to be somewhat repairable. The channels could be cleared and
some minor reshaping could be done with a small razor blade. Therefore, the decision was initially

26

made to not reprint the gears, as


there

was

no

other

viable

alternative. It quickly became


obvious that additional gears
would be needed, since the
tenuous contact between the gears
would skip and wear down the
delicate teeth. When the time came
to print replacements, it turned
out that the previous round had

Figure 18: Inner Gimbal Ring Interface

not been printed at the MakerBots true 100 micron resolution: they had been printed using a lower
setting to speed up production. The replacement gears were therefore printed at the highest possible
resolution, and were of a much higher quality then their predecessors, as expected. They still required
some very minor cleaning up, but resulted in a capable, functional gearing system. The refined gears
were installed and the inner gimbal ring slippage ceased to be an issue.
To further increase gear reliability, future iterations of the gear could feature a design with half
the current number of teeth. This would create a more reliable gear while also preserving functionality.
Mechanical Fastener Selection
Holding the various components of a GAVN module together requires some type of fastener;
since using permanent adhesive prevents the replacement of damaged parts, mechanical fasteners are
the logical choice. Socket head cap screws (SHCS) have cylindrical heads two times the diameter of
the shaft, giving maximum holding area with minimum weight for all sizes. The hex socket transfers
torque well and in most cases is good at avoiding stripping. Using a bolt and nut fastening system also
allows for adjustable holding pressure.
While a standard nut-and-bolt system relies primarily on the compressive force of the system
to fasten parts together, potential problems with this method were discovered during initial fan testing.
The fans selected for the prototype aircraft produce a fairly intense vibration; this vibration was
measured at various fan speeds with a three axis accelerometer recording device (specifically, a
smartphone app) and analyzed, revealing that the nuts could potentially shake lose as a result of the
motor vibration. This issue was resolved by reducing the through-hole diameter of the printed parts
such that the threads of the bolt would bite into the plastic and form a threaded hole. After this

27

revision, the nuts only share part of the holding pressure, as most of the holding force now comes
from the friction between the PLA and the bolt. Loctite thread locker was used to hold the nuts in
place as a backup, in case the PLA failed; further testing showed that the Loctite was able to prevent
the nuts from loosening even during prolonged exposure to the most powerful vibrations.
The material selection of the bolts and nuts was driven primarily by cost and weight. Nylon
bolts have a very low density and could easily handle the loads; however, nylon is not much harder
than PLA, meaning the friction between the bolt and the PLA would be much less than if the bolts
were made of steel. Tool steel was the strongest material available, and therefore best able to withstand
the torque applied during the initial driving process (due to the cutting and deforming of PLA), but
was more expensive than stainless steel by a factor of 1.6. The overall lowest cost, lowest weight, and
highest strength option was 18-8 stainless steel: its high Youngs Modulus allowed the bolts to be
smaller, which aligned well with the overall design goal of space conservation.
In total, only two kinds of bolts and matching nuts were required for the entire GAVN. Two
2mm SHCSs and nuts are used on each of the forward bearing blocks for the 1st stage bearings, and
three of the same bolts (and matching nuts) are used for both the 1st stage servomotor mount and the
counterweight for the 1st stage. Three more of the 2mm bolts and nuts are used to hold the ducted
fan into the 1st stage gimbal ring. A total of six 4mm SCHSs were used on the second stage servo
mount. In total, the weight of all of the mechanical fasters came out to 31.5g per GAVN, less than
the initial weight allocation of 50g per GAVN.
Finite Element Analysis
In order to maximize strength while minimizing weight in the GAVN system, finite element
analysis was used to refine the design. Solid Edge ST5s included FEA package (similar to standalone
stress analysis programs like ANSYS or COMSOL) was used to remove the time-consuming process
of converting and importing digital models between multiple programs.
The standard workflow for FEA software is as follows:
1. Define the geometry.
2. Define the type of analysis (time variant, steady state, etc.)
3. Define the materials and their properties.
4. Define the physics (constraints, fields, forces, fluxes, velocities, etc.)
5. Define the mesh.
6. Solve the model and output the results.

28

A significant number of assumptions were made in step 3 due to the idiosyncrasies inherent
to 3D printing: namely, while the material properties of solid PLA are readily available from various
databases online (the values for the specific filament sold by MakerBot were unavailable), the internal
honeycomb structure the GAVN parts would be printed with result in anisotropic properties that are
not the same as the database values. In addition, since the exact dimensions and layout of the
honeycomb structure change with each print, obtaining a consistent mathematical model or estimation
of the true material properties is extremely difficult. In the absence of a better solution, it was decided
to estimate the parts as if they were made of solid PLA (estimated by the average of the values available
from Materials Web, or Matweb) and utilize a significant safety factor to ensure proper function of
the final part. The table of relevant properties is below in Table 4.
Density
Ultimate Tensile Stress
Yield Tensile Stress
Elongation at Yield
Youngs Modulus

1.27 g/cc
49.3 MPa
44.7 MPa
90.1 %
3.63 GPa

Table 4: Relevant Average PLA Material Properties

The first analysis modeled the deformation and stress on the inner gimbal ring as a result of
the thrust force from the fan. The maximum thrust force of 850g (8.33N) was applied to the entire
circumference as an evenly distributed load on the top face of the inner gimbal ring, and the ring was
constrained in all six degrees of freedom at the bearing contact points (represented by blue dots in
Figure 19).

Figure 19: Inner Gimbal Ring FEA - Displacement Due to Max Engine Thrust

The rings deformation as seen in Figure 19 is highly amplified: the actual maximum deflection
was calculated to be only 0.14mm. Even after adding 0.5 sigma of tolerance (1 sigma total, or 33%),

29

the deflection range is 0.140.046mm, which is well within the acceptable zone for this application.
The stresses resulting from these forces and constraints can be seen in Figure 54 in Appendix A.
A similar analysis was performed on the outer gimbal ring: the force was again set equal to the
maximum engine thrust and applied to the inner gimbal mounting points, while a full constraint was
placed on the rear shaft where it interfaces with the bearings. The model was run for two different
force orientations, perpendicular and parallel to the rings rotation axis: the former orientation is seen
below in Figure 20.

Figure 20: Outer Gimbal Ring FEA - Von Mises Stresses Due to Max Engine Thrust

It can be seen the highest stresses (as seen in red, with a value of 18.6 MPa) occur at the top
of the shaft near the bearing interface, as is expected. These stresses are less than half of the PLAs
yield stress, indicating the part would not deform significantly during use.
Final Design
After component selection and design verification through FEA, the first prototypes of
custom GAVN components (Mark 1, seen in Figure 55 in Appendix A) were printed. After a small
amount of manual reworking to fit everything together, the parts were remeasured and the models
were adjusted accordingly. The modified parts (Mark 1.5, seen in Figure 21) were reprinted and
reassembled.

30

Figure 21: Rendered Exploded View of GAVN Design Mark 1.5

Figure 22: Fully Assembled GAVN Test Stand

The test setup (consisting of the GAVN Mark 1.5, the acrylic GAVN test stand, and all
necessary electronics) (Figure 22) revealed some other necessary changes, but the Mark 1.5 design was
31

functional overall. With these and other cosmetic adjustments made, a full set of Mark 2 GAVNs
(seen in Figure 23) were produced and used for the rest of the project.

Figure 23: GAVN Mark 2 Design Render

Airframe
Initial Configuration
In order to prove the GAVN concepts viability as an aircraft control system, a prototype
aircraft needed to be constructed and flown. An airframe was required to hold the GAVN units as
well as any other components necessary for flight. As said earlier, it was decided that the prototype
aircraft would be a three unit system and therefore the airframe would be based off a tri-copter design.
The principal components on the aircraft would include a battery, the MyRIO controller, three ESCs,
a small sensor suite, required wiring, and the three GAVN units. All of these components are required
to be onboard for flight, and therefore the airframe must be large and strong enough to be able to
handle their mass.
The first two airframe decisions were made simultaneously: the orientation of the GAVN
units to each other (and therefore on the airframe) and the layout of the components on the airframe.
These decisions were critical to the shape and size of the frame and therefore needed to be determined
before intensive design could occur.

32

Two options were considered for the layout of the GAVN units. The first was a T-shaped
configuration, shown in Figure 24, such that two of the GAVN units would face perpendicular to the
forward direction, while the third would be placed on the aft of the aircraft facing to the rear. This
design had the benefit of only requiring one degree of actuation in each GAVN to move the aircraft
forward and backwards; however, it could have complicated other maneuvers and made stationary
stability potentially more complicated.

Figure 25: Example of T-Shaped Configuration

Figure 24: Example of Y-Shaped Configuration

The second option was to use the symmetric Y-shaped configuration. This layout has the
GAVN units all point outward from the center at 120 degree increments, as shown in Figure 25. This
allows for the system to stabilize easily as the GAVNs are able to counteract each other with less
complex actuations. The drawback of this configuration is that the act of moving forward and
backward requires more complicated motion (both axes) for the gimbals. Since the aircraft is meant
to be as stable as possible and the plan is to perform maneuvers in more than just the forward and
backward directions, the team decided to go with a Y-shaped airframe to improve balance, stability
and symmetry in the aircraft.
The

layout

of the components was an

independent decision from the GAVN units.

The

stipulations on the layout required that that all the


components fit in a small footprint, the connections
between the components were simple, and the center of
gravity of the components was centered on the airframe.

Figure 26: Preliminary Component Layout

33

To do this, a preliminary layout was created with


the sizes of each element accounted for (shown in
Figure 26). This was the base idea that future
iterations would build on to create a final
component layout.
The component layout became more
complex when finding the center of gravity (CG).
In order for a layout to be ideal, the CG of the
components must be as close to centered on the

Figure 27: Oversized Body Frame

airframe as possible to allow for simple


integration into the aircraft dynamics. In order to
accomplish this, each component was weighed
and had its individual center of gravity found.
Then, using the preliminary layout, each
component was placed in the orientation shown.
Finally, using simple force and moment equations
and the center of gravity weights and locations,

Figure 28: Design for GAVN Airframe

the components were spaced properly such that


the center of gravity for all the components together was placed in the center of the layout.
CAD Design
The next step in the airframe design was using the CAD modeling tool Solid Edge to create
the structure. Using the Y-shaped orientation for the GAVN units and the previously determined
component layout, the airframe was designed to accommodate both of these component sets. Using
the models of the GAVN units, the appropriate hole pattern and frame size for mounting the units to
the frame was determined. Next, the GAVN mounting points were laid out into the Y-shaped
configuration such that each unit was 120 degrees from the either two: due to all the GAVN units
being identical and the airframe being symmetrical, the geometric center of the airframe is also its
center of gravity. Using this knowledge, the CG of the airframe was lined up with the CG of the
components to determine the location of the mounting points for each component. Finally, structural
supports were added for rigidity and the design was completed.

34

There were two major iterations of the airframe. The first (Figure 27) was a larger body that
was created as the first iteration of the Y-shaped configuration. The model was oversized due to
overestimation of the frame dimensions needed fit all the required components. The next iteration
was the same shape and design as the larger frame, just smaller. The reduction in size was required for
material conservation and cost as well as the reduction of weight. This second iteration is the current
model for the airframe of the prototype GAVN aircraft and is shown in Figure 28 with the
components attached where they will be on the final craft.
Material Selection
The final airframe decision was the material that would be used. Solid Edge has the ability to
apply material conditions to the model in order to estimate weight, inertial properties, and structural
integrity. There are many options to consider when selecting the material for the airframe; however,
the choice will be one that provides the greatest structural support at the lowest weight.
After evaluating options, including aluminum, acrylic, and 3D printed ABS/PLA, by analyzing
material properties and manufacturing simplicity, aluminum was chosen as the final material. This
created a secondary decision of the specific aluminum alloy, 6061 or 2024: 2024 is stronger for the
same density, but 6061 is less expensive. It was determined that the additional strength and stiffness
of the 2024 aluminum was worth the extra cost.
For the ease of manufacturing the airframe was cut using a CNC mill in a two-step process.
The first step included the milling of the airframe from an acrylic sheet. This initial prototype was
used to verify the mounting hole patterns for the components to the airframe as well as allow for the
wiring system to be tested on a full aircraft chassis. The second step of the process was completed
following the confirmation of the design with the acrylic frame. This step involved milling the frame
a second time, however this time out of the 2024-T3 aluminum sheet. Following this manufacturing,
the components were mounted to the aluminum and testing was able to commence in full.
FEA/Stress Analysis
After selecting aluminum based on its mechanical properties, stress analysis was performed on
the airframe design to verify its structural integrity. Using FEA, the team performed all analysis
required to provide design validation, which showed that an aluminum frame of that design would be
sufficient.

35

Two different analyses were performed


on the airframe. In the first (Figure 29), the
frame was fixed at points where the GAVN
units would be mounted and pulled down from
the center. This analysis showed displacement
of less than 2mm and stresses well within
normal limits.
After assembling the frame and seeing
deflection beyond the predicted 2mm, a second

Figure 29: FEA with GAVN Mounts Fixed

analysis (Figure 30) was performed that instead


fixed the center of the airframe and applied the
forces to the GAVN mounts, allowing them to
rotate as had been seen in the lab. This second
analysis showed more displacement compared
to the first test, up to 18mm, which more
accurately depicted what was seen in testing.
This analysis was determined to prove the
material adequate for our design as the bending
shown was allowable in the aircraft, and

Figure 30: FEA with Airframe Center Fixed

necessary to allow for the aircraft to be as


lightweight as possible.
Electrical System
Power Distribution
While simple in principle, designing the power distribution harness had to be done carefully
due to the severity of consequences if wiring was done improperly (specifically, for bodily harm and
aircraft damage). The power distribution harness had three functions:

To distribute power from the single power source (battery) to the four primary
components of the electrical system.

To allow the myRIO to be powered first and the ESCs second, which was required
for proper initialization.

To allow easy replacement and assembly of components.


36

In the design of the wiring harness the first consideration was voltage. The battery output
voltage ranges between 12 and 16.8 volts, which the ESCs could tolerate (input voltage range between
8 and 24 volts) but was slightly outside the myRIOs range (8 to 16 volts). After consulting the
myRIOs accompanying documentation, no additional information on the voltage tolerance range was
found. The assumption was made that as a consumer electronic device it would not be damaged by
the 16.8V output from the battery; after wiring the two together and doing a functionality check, this
assumption was proven to be correct.
The next consideration in the design of the wiring harness was current capacity. The wire
purchased for the harness was 14 gauge AWG stranded wire, with an ultra-high temperature
polydimethylsiloxane (PDMS or Silicone) sheath, providing it an unusually high maximum continuous
current rating of 50 amps (normal plastic coated wire can only handle 25 amps at 90C, according to
the American Wire Gauge Standards Table). 14 gauge wire was chosen over larger wire because it is
the same wire that comes pre-attached to the ESCs themselves; using larger wire would have added
unnecessary weight to the aircraft. To make changing components easier, HDX barrel-style gold
plated quick-disconnect plugs were selected as appropriate connections between the various
components and the wiring harness.
The HDX connectors are made in such a way that if assembled properly, connecting wires
backwards is impossible. This is critical because connecting an ESC to the battery in reverse can have
dangerous and disastrous consequences. During the winter term at an open house demonstration,
before the HDX connectors were acquired, an ESC was connected backwards to the battery, resulting
in an immediate surge current with toxic smoke release and wire fusing. The situation was resolved
before more serious damage occurred, but the ESC was completely destroyed, and the battery
connector was reparably damaged. If this situation wasnt resolved as quickly as it was, the ESC could
have caught fire, and the battery may have ruptured and caught fire. The HDX connectors are an
intelligent engineering safety mechanism that provides a physical safety system preventing damage to
personnel and components.
Notably absent from the power diagram are the six actuation servos, which operate at 6 volts,
well below the range of the main harness. The initial power distribution harness included a battery
eliminator circuit (or BEC), which would step down the voltage with a transistor-transistor circuit and
feed it to all six servos via a custom power distribution block. After purchasing the components
necessary and testing the circuit, the ESCs were discovered to have small integrated BECs that could

37

power their GAVNs servomotors. As such, the external BEC was eliminated from the system; the
servomotor power connections appear on the signal harness diagram in the next section.
In total, the wiring harness consists of black and red sheathed 14 gauge Turnigy wire, two
sets of 90 amp, 4.5mm HDX connectors, and three sets of 40 amp, 3.5mm HDX connectors. Below
in [Figure 1], the power wiring diagram is shown, indicating connectors (green) and all the power
components. All of the components in the circuit are in parallel to the battery, and thus operate at the
same voltage, but varying amperages, depending on that devices current demand.

Figure 31: Power Distribution Wiring Schematic

Signal Wiring
The GAVN prototype electrical system is diagrammed below in Figure 32. The six
servomotors, three ducted fans, three ESCs, battery, battery monitor, and myRIO are all shown. As
explained above, each BEC powers the two servomotors for its GAVN. Each servomotor has a three
wire connection (brown negative, red positive, and orange signal). Custom connectors were
fabricated to distribute the power from the one BEC output per ESC to the two servomotors.

38

Figure 32: GAVN Prototype Signal Wiring Harness

The battery monitor was included as an inexpensive yet critical safety device. Its LED
display and two buzzers (both output a 500Hz tone, one at 90dB and the other at 110 dB, so as to
be audible over loud engines) display each battery cells voltage as well as the total voltage. The
small, 1 square device (of negligible weight) connects to the batterys balance charging port, from
which it reads the battery cells power levels and sets off an alarm if the voltage of any one cell drops
below a user-set cutoff. If a cell drops too far below its designed operating voltage (for this
prototype, the cells are designed to operate at 3.30V, with a failure point of 2.9V), it may fail and
cause a thermal runaway fire.

39

Prototyping
GAVN Component List
The following component list (Figure 33) includes all the components that will be used on the
prototype aircraft. All are actual weights except for the airframe (estimated in modeling software) and
wires; the optical flow sensor and 9 degree-of-freedom positioning sensor are not included either, but
their weights are negligible.
Component name/type
5mm ID bearing
10mm ID bearing
Bearing block clamp
Inner gimbal ring
Inner ring servo mount
Counter-weight
Outer Gimbal Ring
Gimbal Components
Outer ring mounting block
Small Gear - Bore
Small Gear - Servo
Large Gear - Bore
Large Gear - Servo
Electric ducted fan A
Electric ducted fan B
Motors
Electric ducted fan C
Electric ducted fan D
Servos
BMS-385DMAX Servo (w/ screws)
2mm SHCS
Mounting Hardware 2mm ID nuts
4mm SHCS
MyRIO Board
MyRIO
MyRIO Case
ESC
Turnigy Plush 40A ESC
1-8s Lipo Battery Tester/Buzzer
Battery
Battery
Total airframe
Airframe
Wires
Bearings

Weight (each)
0.375
5.2
6
10
10.9
31
14
18.6
0.7
0.7
1.4
1.6
1.0
70.4
70.7
70.3
18.3
0.5
0.1
2
92
100
36
9.2
560
156
60

Per GAVN
Qty Weight
2
0.75
2
10.4
2
12
1
10
1
10.9
1
31
1
14
1
18.6
1
0.7
1
0.7
1
1.4
1
1.6
0
0
0
0
0
0
0
0
2
36.6
10
5
10
1
5
10
0
0
0
0
1
36
0
0
0
0
0
0
0
0

Whole Craft
Qty Weight
6
2.25
6
31.2
6
36
3
30
3
32.7
3
93
3
42
3
55.8
3
2.1
3
2.1
3
4.2
3
4.8
1
1.0
1
70.4
1
70.7
0
0
6
109.8
30
15
30
3
15
30
1
92
1
100
3
108
1
9.2
1
560
1
156
1
60

Figure 33: GAVN Component List

All COTS components and their supplier (with the exception of the airframe, which has not
been decided yet, and standard components like mounting hardware and wires) have been included in
Table 5.

40

Category
Microcontroller
Servo
Ducted Fan
Electronic Speed Controller
Battery

Description
NI MyRIO
BMS-385DMAX Digital Servo
AEO 64mm 5250KV Brushless
Turnigy Plush 40A ESC
Turnigy nano-tech 5000mAh 4S 35~70C Lipo Pack

Battery Case

Turnigy Soft Silicone Lipo Battery Protector (Black)


155x52x38.5mm
HobbyKing.com

Optical Flow Sensor

PX4Flow Kit

Combination Accelerometer/
Gyroscope
9 Degrees of Freedom MPU-9150 Breakout

Source
NI
HobbyKing.com
RCEcho.com
HobbyKing.com
HobbyKing.com

3D Robotics
SparkFun

Table 5: GAVN COTS Components & Sources

Center of Gravity Testing


Obtaining an accurate measurement of the center of gravity of the components on the aircraft
was critical to optimizing the design of the GAVN and the aircraft. Each component had its CG
determined using the same method; the example below explains the procedure as used for determining
for the ducted fans center of gravity.
Precisely locating the thrust sources center of gravity at the GAVNs actuation center is
important, as it eliminates any gravitational moments from the gimbal system and thereby reduces the
number of elements needed to be considered during dynamics calculations. In order to obtain this
measurement, the team utilized (with slight modifications) a technique cited in both SAE conference
papers [11] and Department of Defense Test Operation Procedures.
The test is fundamentally simple:
1. Hang the fan from any point on its body, ensuring that it is free to rotate about the
attachment point (i.e. not fixed at a certain angle, or inhibited from rotating in any way).
2. Use a plumb bob (or other dense object) and string to mark a vertical line, originating at the
attachment point, on the object or on paper temporarily affixed to it; this line will pass
through the center of gravity, as it will have settled directly beneath the attachment point.
3. Repeat steps two and three, hanging the object from a different point on the body that will
still allow the tester to mark the vertical line on the same face of the object (or paper).
4. The point where the two lines intersect is the center of gravity of the object.

41

Figure 34: Ducted Fan Hanging with Weight Suspended

Figure 35: Ducted Fan Image Annotated with Measurements

The original test procedure also utilizes strain gauges and/or portable wheel scales to
numerically calculate the longitudinal position, but this was unnecessary for the ducted fan given its
axisymmetric shape. The team measured the axial position of the center of gravity and incorporated
it into the physical properties of the CAD model, allowing the center of gravity to be precisely aligned
with the gimbal center of actuation.
3D Printing
All of the custom-designed parts were fabricated by a 3D printer. Drexel University owns and
operates a MakerBot 3D Replicator 2 and made it available for rapid prototyping of the GAVNs
42

components. 3D printing has many advantages, including but not limited to: fast iterations, allowing
designers to verify small design changes and catch flaws sooner, keeping costs down; parts are made
of a material that is tolerant to hand-tool use for touching up complex shapes; and low turn-around
time as compared to off-site fabrication companies. The MakerBot Replicator 2 utilizes fused filament
fabrication technology coupled with polylactic acid (PLA) plastic filament to produce 3D printed parts
with a resolution of as small as 100 microns. The MakerBot software can simultaneously produce as
many parts as can be arranged on the printing stage. Access to the MakerBot has been a significant
asset for the prototyping phase of the GAVNs design for these reasons; however, it was not without
its share of headaches.
Due to the fact that the MakerBot extrudes onto the printing stage, any parts that include an
overhang need to be printed with support posts that the MakerBot can then use to prop up the
complex geometry of the part being printed. This requirement has complicated the production of
several of the GAVNs components since the inner and outer gimbal rings each contain posts that fit
into the ball bearings. In one instance, the post that secures the outer gimbal ring to the servo
mounting block was warped and the hole for the bolt was not fully cleared. The post was the sanded
down with hand tools to fit the bearing, and the bolt hole was drilled out to accept the bolt. The
warped shape of the post caused it to not fully fit the hole in the bearing, resulting in a wobble unless
the bolt was fully tightened. While this flaw was not severe enough as to hinder prototype assembly,
it would not be suitable for a final assembly. In another instance, the inner rings two axles came out
skewed and warped; the most effective method to mitigate this behavior is currently under
investigation.
In addition, the PLA material has been discovered to be quite sensitive to heat. A heat gun
was used during a wiring test to heat shrink tubing onto a soldered wire junction that passed over the
vertex of the outer gimbal ring. It was realized after the fact that the heat had softened and distorted
the parts surrounding the wire junction. The affected parts needed to be re-printed.
A well-known drawback to the use of 3D printers such as the MakerBot is the shrinkage of
the material as it cools. All construction materials have a coefficient of thermal expansion that
describes the materials change in volume due to changes in ambient temperature, and printing
mediums (which are heated during printing) such as PLA are engineered to resist these changes to
conserve true design dimensions and ensure print accuracy. For the majority of the printed parts, this
small degree of shrinkage was not a significant issue. It has only posed an issue for the smallest features

43

of the design, namely the hexagonal counter-sunk nut holes. The printed holes frequently require
cleaning up with a blade to remove unwanted filler material and refine corners to ensure a proper fit.
Once the full assembly began initial testing, the need quickly arose to keep backup parts in
stock. Over the course of numerous component revisions, several parts became worn out or broken
entirely from the stresses of component assembly and fatigue from repeated loading cycles. The lead
time for replacements could be up to 3 days, so it made sense to keep a small backlog of replacement
parts on hand. The GAVNs modular design made swapping out parts fairly straightforward and
facilitated simple part replacement procedures.
Finally, while the printed PLA material is lightweight and strong, and flexible enough to preload the axles and bearings as discussed above, it is somewhat brittle and will snap if subjected to
substantial loads. Though this can be frustrating, it does help ensure that easily replaceable parts will
break first during a crash, instead of more valuable components. The outer gimbal ring in particular
was the most delicate. Due to its long, relatively thin features, it was susceptible to snapping if proper
care was not exercised during testing and assembly. This made the outer gimbal ring the most often
replaced part.
Thrust and Voltage Testing
The most important component of any aircraft is its propulsion. The ducted fans that were
selected for the GAVN prototype needed to be benchmarked and tested to establish several
operational parameters, chiefly among them the potential thrust. The manufacturer cited the
maximum thrust output to be 670g of thrust (a theoretical thrust-to-weight ratio of 9.2:1 given the
fans weight of 73g), however no additional thrust specifications were provided.
In order to characterize the power input to thrust output relationship, a test was designed to
measure the thrust output of the engine over a range of input voltages. A stand was constructed to
secure the ducted fan onto a digital scale in such a way that the fan thrusts downward into the scale.
To measure the voltage, voltmeter leads were affixed to the wires supplying power to the fan from the
ESC, and the thrust was controlled via a CCM controller. To conduct a test, the digital scale readout
and the voltmeter readout are positioned such that they can be simultaneously filmed by one camera.
During each test run (with cameras running) the throttle was gradually increased until the engine
reached 850g of thrust. Examining the test footage frame by frame allowed accurate voltage-thrust
input-output pairs to be recorded, as seen below in Figure 36. Full test results are available in Appendix
B: Thrust Test Data.

44

.
800
700
Test 1
Test 3
Test 4
Test 5
Test 7
Test 8
Test 10

Thrust (g)

600
500
400
300
200
100
0
1

5
6
RMS Voltage (V)

10

Figure 36: Thrust-Voltage Data for Fan B

The test stand was initially constructed from bass and balsa wood and held together with wood
glue, but it rapidly became apparent that the ducted fan needed to be more strongly affixed to the test
stand, as the frictional force on the fan duct was not enough to hold it in place. To provide additional
support, two metal rods were inserted into the sides of the test stand such that the top edge of the
engine cowling rests atop the two rods. This successfully prevented the fan from thrusting out of the
frame.
Over the course of numerous thrust tests, the vibration from the motor caused the structural
integrity of the wooden stand to suffer, and the stand eventually needed to be replaced. The secondgeneration test stand maintained the layout of its predecessor, but was improved with stronger
material, computer-aided design, and superior construction. The stand was engineered to resist vertical
movement of the engine, and also to inhibit rotation of the engine within the stand due to adverse
torque. The new stand was made up of 3 laser-cut plates of acrylic built into a 2-walled structure with
a top plate that held the ducted fan. The plate junctions are mated with dovetail joints and bonded
together with a cyanoacrylate adhesive. The laser-cutter pattern used to cut the acrylic plates is shown
in Figure 37.

45

Figure 37: Mark 2 Test Stand Pattern

GAVN Test Stand Design & Fabrication


Once the GAVN prototype was assembled, there was a need for a new stand that would
support the GAVN unit during testing and maintenance. The GAVN had to be elevated during testing
so as to not hinder movement of the primary or secondary servo rings, and to keep it anchored during
thrust testing. Much like the thrust test stand, the GAVN support stand needed to be solid enough to
resist the forces and rocking motions of the GAVN during operation, and was therefore constructed
of laser-cut acrylic. The design of the stand (as seen in Figure 38) centered around 2 vertical supports
that would slot together to form an X-shaped cross section, which was is in turn seated into a lower
plate that becomes the foot of the stand, and an upper plate that serves as the GAVNs mounting
surface.
The test stand prototype met all the demands that were required of it, and did not require any
additional redesigns.

Figure 38: GAVN Test Stand Pattern

Control System Component Testing


The critical components of the tricopter system that are controlled are the electric ducted fans
and the servos that actuate them. After taking delivery of these components, the immediate goal was
to confirm their compatibility with the myRIO. Since both the fans and servos are widely used in R/C
aircraft hobbyist communities, their control methods are standardized to use pulse width modulation
46

(PWM) signals. Both use a 50Hz wave with a 5% to 10% duty cycle. For the fans, this duty cycle range
corresponds to minimum and maximum thrust. For the servos, 5% duty cycle corresponds to a 0
angle, while 10% turns the servo shaft 180. The myRIO contains a total of 8 digital I/O pins that
possess a secondary function of PWM generation, as shown in Figure 39. These pins can be utilized
by LabVIEWs PWM Express VI.

Figure 39: NI myRIO Connection Diagram; MXP-A & -B (left), MSP-C (right)

The servomotors have a standardized 3-wire connection, shown in Figure 40. Several
variations of servo wire coloring exist, but the wires can be identified by a simple rule-of-thumb: wires
are always ordered ground-power-signal and ground is the darkest colored wire. During early servo
testing it was discovered that the myRIO cannot provide enough current to power the servos: it can
only provide a maximum 100 mA at 5V on each connector. Testing the
servo under no load with an Arduino and an in-line multimeter showed
that the servo draws about 200 mA when moving. Introducing some
resistance to impede the servos motion caused its current draw to spike
to about 400 mA. Therefore, in order to control the servo with the myRIO,
an external source must be used to power the servo. For testing purposes,
a smartphone USB wall charger was modified to utilize its power and

Figure 40: Standard Servo


Wire

ground wires: the charger was chosen for its convenient output of 1A at 5V, which is sufficient to
power two servos simultaneously. When using an external power source to power a component, the
ground of that source must also be connected to one of the myRIOs ground pins.
The throttle of the electric ducted fan is controlled in an identical fashion. An electric speed
controller (ESC) connects the two leads from the battery to the three leads of the fan, and converts

47

DC power from the battery into the three-phase AC power that drives the brushless motor. The ESC
also has a servo wire which controls the speed of the fan. Similar to the servomotor connections, the
ground of the ESC servo wire must be connected to a ground on the myRIO and the signal wire
connected to a PWM pin on the myRIO.
A LabVIEW VI was created to test the operation of the servos and fans. Only one VI was
made since both components function via the same principle: PWM. The block diagram (Figure 41)
consists of three parallel while loops, each serving a different purpose.
The first loop (left) increments the PWM duty cycle by 0.001 (0.1%) every five seconds,
starting at the minimum duty cycle of 5%. This is accomplished using a for loop with shift registers.
The maximum duty cycle is designated by setting maximum number of loops for the for loop. The
for loop is also contained within a case structure which acts as an if statement, controlled by a Boolean
switch. While the switch is off (false), the duty cycle increases incrementally as described. However,
when the switch is on (true), the for loop is disabled and the duty cycle remains at 5%. This option
exists because the ESC and motor must first initialize by receiving a minimum throttle signal (5%)
when powered on before it can operate normally. The incrementing duty cycle value is saved to a local
variable called RT Duty Cycle, which can be accessed by other loops within the VI.
The second while loop (top right) designates the stop conditions for the VI. It causes the entire
VI to stop running when either the for loop finishes, or if the stop button on the front panel is pressed.
The duty cycle is also reset to 5% when the stop conditions are met.
The third loop (bottom right) communicates with the myRIO. It generates a 50Hz PWM
signal on the desired DIO (digital input-output) pin. Duty cycle can be set by the local variable RT
Duty Cycle or controlled manually with a slider. This loop also reads the PWM signal on an analog
input pin and displays it on a chart.

48

Figure 41: MotorTest VI Block Diagram (top) and Front Panel (bottom)

Two sensors were also purchased to measure the current state of the tricopter: an inertial
measurement unit (IMU) and an optical flow camera. The MPU-9150 IMU features a three-axis
gyroscope, three-axis accelerometer, and three-axis compass. The PX4Flow optical flow camera is
used for measuring translational motion using on-board image processing. It also measures altitude
using a built-in ultrasonic senor. These sensors would have been sufficient; however, integrating them
into the tricopter system proved difficult. Although both sensors were confirmed to be functional, a
simple and straightforward method for connecting to the myRIO and extracting meaningful, useable
data was not found. The sensors support various serial protocols for communicating data (I2C, UART,
USB), but utilizing them required comprehensive knowledge of computer science beyond the abilities
of the development team.

49

Code Development
Developing code for the GAVN tricopter proved to be a very iterative process. Not only did
the absence of position sensors make any kind of autonomous operation impossible, but the flight
dynamics control system also could not be implemented. The control methods described in the
previous section require a more thorough understanding of LabVIEW and real-time targets. In lieu of
this, a manual control scheme was created.
Since the myRIO has eight dedicated PWM pins spread across its three connector ports and
nine PWM signals are needed, the first priority was to create enough PWM outputs. This was possible
using the myRIOs field programmable gate array (FPGA) chip. The FPGA personality was modified
so that nine PWM outputs were created on the MXP-A connector, part of which is shown in Figure
1. The custom FPGA personality has PWM outputs on DIO 0 though DIO 8, and produces PWM
signals by controlling its period and pulse width in microseconds. It is important to remember that
while utilizing this modified FPGA personality, none of the secondary pin functions will work. For
example, as shown in Figure 34, Pin 27/DIO8 could not be used to create a PWM signal using the
PWM Express VI.

Figure 42: FPGA Personality for PWM Outputs

The LabVIEW project built to control the tricopter consists of two main components:
KeyboardControl.vi on the host PC, and RT Main.vi on the real-time target (the myRIO). The
KeyboardControl VI assigns functions to various buttons on the PCs keyboard. This is achieved
using an event structure and a case structure, as seen in Figure 2. The event structure allows the user
to declare what happens when a key is pressed and when it is released, while the case structure allows

50

different functions to be prescribed to different key presses. The KeyboardControl VI controls three
separate PWM signal: one for the outer gimbal rings, one for the inner gimbal rings, and one for the
fans. The respective components of each GAVN unit are receiving the same signal, i.e. all three inner
rings are controlled simultaneously with one signal. The up/down arrow keys increase/decrease the
pulse width of the signal to the inner rings, the left/right arrow keys control the outer rings, and the
W and S keys control the fans. Two emergency off switches were also created for the fans: when the
number pad enter key is pressed, the fan pulse width returns to its minimum and stops the fans from
spinning; the space bar will execute a more extreme emergency off by resetting all PWM signals to
their initial values before stopping all VIs. Sub VIs were also created to limit the pulse values to within
their proper ranges. The servos are restricted to 800-2200 s (4-11% duty cycle). The fans are limited
to a range of 50-100; this value is scaled up by a factor of 20 to represent the pulse width in
microseconds. These signals are stored in a cluster that is written to a network-published shared
variable called KeyData, which is used to communicate data to other VIs running in the same
network. The RT Main, which runs on the myRIO, receives the KeyData shared variable and writes
the pulse width values to the appropriate pins designated in the custom FPGA personality.

Figure 43: KeyboardControl.vi Block Diagram

A second host VI was created to demonstrate two-dimensional directional control of the


tricopter. This VI, shown in Figure 3, was expanded from the KeyboardControl VI to control
individual signals for each servo and one signal for all three fans. The logic behind this code is to
orient each fan to the necessary position in order to cause the tricopter for move in a particular
direction. For example, when the up arrow is pressed, all servos will move the fans so that the vehicle
will move forward. They fans would remain in this orientation as long as the key held down. When
the key is released, all fans return to a vertical position.

51

Figure 44: KeyboardControl_v2.vi Block Diagram

A trim functionality was also added to the real-time VI, as shown in Figure 4. These controls
allow for minute, independent adjustments to each servo and fan, which are performed on the signals
output from the host VI before they are written to the myRIO pins. Such adjustments are beneficial
to the GAVN system since they can help compensate for any manufacturing or assembly discrepancies
in the system design.

Figure 45: RT Main.vi Block Diagram

Code development was a very iterative process. The code was constantly being updated and
improved based on the lessons learned about the nuances of LabVIEW. The biggest obstacle was
52

properly communicating commands to the myRIO. The myRIO is capable of connecting directly to
a computer by creating an ad-hoc WiFi connection. A shared variable was selected as the most logical
means of communicating between the host VI and real-time VI. However, a significant amount of
troubleshooting was needed before the shared variable was operational. Windows Firewall and IP
addresses were impedances to communication. When the project is first opened, LabVIEW
automatically records the IP address of the host and the real-time target specific to the network to
which they are connected at the time. Switching between networks causes the incorrect IP address to
be logged, rendering communication impossible. Connecting to the myRIOs ad-hoc network before
open the LabVIEW project solves this issue. Components of Windows Firewall blocked
communication with the myRIO; disabling the firewall allowed for a proper connection and eliminated
all errors in LabVIEW.
A second control input method of using a Sony PS3 controller was also considered. The PS3
controller supports Bluetooth wireless, which it would use to connect to the host PC. A VI from the
LabVIEW community example site was used to confirm that the PS3 can be recognized in LabVIEW.
However, errors arise when trying to integrate it into the GAVN LabVIEW project. This issue remains
unresolved and was abandoned in favor of keyboard control inputs since that method was already
confirmed to function properly.

Performance Testing
Once the prototype aircraft was fully assembled, a preliminary systems check was conducted.
Using the fundamental code that had been written alongside the assembly of the hardware that it was
intended to control, both the servos and the fans could be controlled wirelessly via the control
software on the computer by this final stage of construction.
Due to the fact that there had been no formal stability testing done to this point, extra steps
were taken to ensure the safety of the craft as well as the safety of the user. To accomplish this goal,
the craft needed to be secured in some way so as to prevent it from losing control and potentially
crashing into the floor or surrounding objects. The simplest solution was to secure each corner of the
craft to the floor with a short length of cord. While the floor of the test space could not be modified
to accommodate lash posts, an alternative solution was devised that consisted of securing the
prototype to the outmoded acrylic frame with nylon cord. The acrylic frame was then weighed down
with 20 pounds of steel plates as ballast. This test configuration can be seen below in Figure 46.

53

Although this method


of securing the aircraft does
prevent the craft from flying
away from the tester and
colliding with nearby objects,
possibly even causing bodily
harm, the restraining nylon
cord has no effect on the
crafts lateral stability. As
such, the craft can strafe
freely from side to side, and
pitch nearly upside down if it
is allowed. Until the crafts

Figure 46: Prototype Aircraft Test Setup

stability is suitably refined, a tester was and will be required to support and guide the craft as it throttles
up and achieves lift. During the first several rounds of testing, the aircraft was very unstable and
required a firm grasp on the part of the operator to keep it from crashing. After each test, the alignment
of the GAVNs relative to one another was adjusted, and as a result the stability was very gradually
improved. Due to time restrictions, fine-tuning the balance of the craft was still ongoing at the time
of this writing. During the last several tests, the craft could be balanced and supported by only one
finger, albeit only for a few seconds at a time.
It was noticed during testing that when the motors ran above a 70% PWM duty cycle for a
period of more than a few seconds, they began to lose thrust in a random manner. After
troubleshooting, it was initially thought that the issue was most likely related to maximum available
current, as the engines draw nearly the battery's rated maximum continuous current (90 amperes vs.
100 amperes) during hovering flight. In an attempt to quickly solve this issue without buying a more
capable battery, a simple wire harness was made such that the craft could be tethered to and powered
by both batteries in parallel, creating a maximum available current of nearly 200 amperes while
removing a significant amount of weight from the aircraft, allowing for flight at much lower thrust
levels. A five foot long tether cable was made to transmit power from the battery pack to the aircraft.
Testing continued with this new battery setup, but the fan power issue persisted. The only conclusion
remaining is that the batterys voltage is not high enough, and as voltage drops during discharge, it
quickly falls below the required voltage for the fans to maintain that thrust level. The solution is to
54

replace the current battery with one of a higher voltage: each lithium-polymer cell has about four volts,
and thus a 5S battery with a nominal 18.4 volt rating may solve the issue. This solution would likely
create a new issue for powering the myRIO, which (as explained earlier) is rated for a maximum input
of 16 volts. A step down power circuit such as a buck converter would be needed in future iterations.
After confirming the hovering capability of the aircraft, the other principal goal of longitudinal
motion needed to be tested as well. To accomplish this, the aircraft had its landing legs covered with
low-friction Teflon tape, and was set on a tile floor. The craft-floor interface had a sufficiently low
coefficient of friction so as to allow the craft to slide across the floor with even a gentle push. The
effect of the remaining friction was compensated for by removing the battery from the aircraft, thereby
reducing its weight and frictional interaction and powering the craft by tether, as explained above.
During test runs, the fans were throttled up to a speed that allowed them to produce some force, but
not enough to lift the craft off the ground. The fans were then gimbaled in each direction, and the
craft was allowed to strafe along the floor. This test was immediately successful. The GAVN units
operated smoothly and with impressive synchrony. The movements of the craft were precise,
deliberate, and smooth.
While these tests were rudimentary and rather unstructured, they clearly demonstrated the
desired capability of the aircraft: the craft had shown significant promise in its ability to maneuver and
steer itself. The hover mode needs ongoing refinement, but will undoubtedly achieve total balance
with additional fine-tuning. While additional work will improve the capabilities of the prototype, the
tests conducted this year demonstrate the viability of constant-magnitude thrust vectoring through
showing the ability to maintain level flight during rotational and translational maneuvers in a
flexible platform for various existing and future engine types, fulfilling the three primary needs that
were able to be tested.

Project Impacts
Accurately assessing the downstream impacts of a project during the early phases of project
development is a challenge for concepts that will be delivered to market as a finished product;
developing a design idea that is intended to be flexible and useful in a wide variety of mission profiles
is even more difficult. Analysis of the possible economic, social, ethical, and environmental impacts
of this design project follow, with the caveat that they are strictly limited to the concept and technology
itself, and do not consider or include the impacts of the implementations themselves.

55

Economic and Social


The economic and social impacts of this design project appear to be positive overall. The
versatile design augments current propulsion methods rather than antiquate them; in fact, additional
propulsion methods may become feasible for VTOL aircraft, which would increase business and
employment opportunities. The GAVN design should be accessible to most companies interesting it
by virtue of it being a new implementation of existing technologies: no new components or
manufacturing processes are anticipated to be required. Implementation by the general population
would likely be facilitated by companies who build and sell GAVNs to consumers. Direct impacts to
the general population beyond what has been discussed to this point are minimal: if any significant
impacts were found they would be the result of the implementation, and not the result of the GAVN
technology directly.

Environmental and Ethical


Analyzing the environmental impacts of the GAVN design are especially difficult, and again
appear to be governed more by the implementation. As discussed above, the design does not utilize
any unique materials that are not already common in the aerospace industry, and therefore will not
require and novel recycling or other end-of-life procedures. The durability and lifespan of a GAVN
will be determined by the use case and quality of construction and materials, and therefore cannot be
realistically analyzed at this time.
Ethical consequences of the GAVN design are even more nebulous. Possible civil uses of
GAVN-equipped aircraft include infrastructure inspection, which would be considered ethically
positive in that it can detect structural damage sooner or more accurately, and possibly save lives as a
result of earlier warning. Humanitarian missions like disaster relief aid delivery or search and rescue
operations are also decidedly positive; the improved maneuverability of GAVN-equipped aircraft
could greatly improve the efficiency and success of these missions. Finally, defense implementations,
the likely first step for GAVN technology, could range from supply delivery to surveillance to direct
fire missions. These possibilities are ethically neutral at best, and certainly at least ethically ambiguous.

56

Project Management
Team Organization
The team was organized such that each member was responsible for the portions of the project
that best fit their interests and skills while also remaining able to contribute where they see fit. The
project was conceived by Matthew Brenner during the summer of 2013. Dr. Ajmal Yousuff signed on
as faculty advisor for the project and the team was formed prior to the start of term. Along with being
the originator of the project, Matthew also took charge of researching the projects background and
laying out its principal needs. Graham Donaldson was the organizational and editorial lead of the
team, assuring that all members were aware of meeting times, submitting weekly status reports to both
the senior design program and Dr. Yousuff, and taking responsibility for primary editing of all
submitted works. Christopher Killian was in charge of concept development for the project, making
sure the project remained on track and proper aspects of the development were focused on. Henry
Tarplin and Robert Walto worked together to take the lead on the exceedingly difficult task of the
dynamics and controls aspects of the project.
As development proceeded from theoretical work to construction and testing, some roles
shifted. Matthew and Christopher took on mechanical design responsibility: Matt created the GAVN
and did FEA analysis, while Christopher focused on the airframe. Henry shifted his focus to electrical
systems and design testing, and Robert was the primary programmer, with debugging assistance from
Christopher. Graham remained the organizational and editorial lead, while taking over project
management responsibility from Christopher.
Various organizational tools were implemented to ensure a smooth development progression
for the project. A Google Group mailing list was created to simplify group-wide communication and
sharing of useful information or documents. Google Drive allowed for members to easily collaborate
on reports and documents concerning the team. In order to facilitate the creation of project milestones
and ensure that the project remained on schedule, a Gantt chart was used and will be elaborated on
more in the following section. Team meetings, both with and without Dr. Yousuff, were held regularly
to keep the group organized and work on development together.

Schedule and Milestones


A Gantt chart was created and utilized to create project milestones and track the progress
throughout the duration of the project. Major milestones laid out in the chart coincide closely with
the elements of the phase-gate model, as outlined previously. Phase 0 was completed prior to the start
57

of term and finished during the last weeks leading up to the term, after the team was finalized. Phase
1 began prior to the term and continued until Week 2 to allow for proper formation of the scope of
the project.
The majority of the fall term consisted of the Phase 2 work. All goals for the fall term involved
defining the scope of the project, starting with clearly defining the deliverables (with the assistance of
Dr. Yousuff). Following weeks were spent meticulously studying the needs and specifications to
further develop the concept. Halfway through the term, after urging from Dr. Yousuff, a project
abstract was submitted and accepted to present at the American Institute of Aeronautics and
Astronautics Region 1s 2013 Young Professionals, Students, and Educators Conference (AIAAYPSE) at Johns Hopkins University Applied Physics Lab in Laurel, Maryland on November 15th, 2013.
After learning that the abstract had been accepted, the milestones for weeks 6-8 were restructured so
the team would be prepared to present at the end of week 8.
Phase 3 of the model was primarily completed over winter term. Some work on Phase 3 was
begun during fall term so that team members could develop adequate technical knowledge of the
dynamics and control of the system in time to present at the AIAA-YPSE conference, however all
design and technical work was completed during the winter. The goals for the term were all related to
the detailed design of the GAVN system. At the start of the winter term analysis of COTS components
required for the project began, as they needed to be ordered and received prior to further design. The
main goals of the term were to complete designs for the GAVN unit and prototype airframe. The
GAVN unit design went through multiple iterations throughout the term and was completed on
schedule. The airframe design suffered some early setbacks and at the conclusion of the winter term
was approximately ninety percent complete. At the start of the spring term the airframe design was
analyzed and completed by the end of week 1. By the end of week 4 manufacturing was completed
and therefore Phase 3 was complete.
Phase 4 was completed during the spring term of senior design. This phase consisted of the
final assembly of the aircraft as well as all the testing done for the project. Initial assembly took one
week, including mounting all components and wiring for all systems. The testing took place between
weeks 4 through 7 of the term. The tests began with general systems tests, proving that the servos and
motors worked properly and in a cohesive system, this was achieved by the end of week 4. Weeks 5
through 7 consisted of various flight tests. These tests included hover and 2D motion. All tests were
completed by the end of week 7. The presentation at the start of week 8 signified the end of work on

58

the project for this senior design year; however, further work is capable in phases 4 and 5. This work
is hoped to be carried out in following years of senior design.

Project Budget
Expenditures on this project to date total $1,342.89. Table 6 summarizes the purchases.
Date

Company

Order #

Items Summary

2014-01-17
2014-01-17
2014-01-17
2014-01-30
2014-01-31
2014-02-05
2014-02-06
2014-02-09
2014-02-11
2014-02-20
2014-03-09
2014-03-10
2014-04-01
2014-04-01
2014-04-07
2014-04-23

HobbyKing
HobbyKing
RC Echo
Amazon
HobbyKing
3D Robotics
Amazon
Sparkfun
McMaster
McMaster
HobbyKing
MakerBot
OnlineMetals
McMaster
3D Robotics
McMaster

2008488136
2008487716
9655
103-7837813-4881046
2008577357
R416146072
103-5231420-4447423
897665
73429534
74729132
2008822599
100086589
907785
79670427
R701407650
82477593

Power supply & charger


Servos, ESCs, Batteries
Engines
Battery alarms
Electrical connections
Optical flow sensor
Earplugs & battery sleeve
Gyro
Bearings, hardware
Screws
Wires & replacements
Ocean Blue PLA (1kg)
Aluminum
Nuts, sleeving, drivers
Sensor Connectors
Screws & washers

Amount
$
$
$
$
$
$
$
$
$
$
$
$
$
$
$
$

143.31
403.05
76.66
15.30
45.75
152.31
29.02
43.88
144.64
27.57
61.06
69.07
63.52
37.69
16.12
13.94

Table 6: GAVN Team Expenditures

An application was also submitted to and approved by National Instruments to receive a


demonstration copy of their new myRIO control system, saving a significant amount of money. The
team also received a generous $500 grant from The Boeing Company, offsetting the expenses paid for
by Dr. Yousuff down to $842.89.

Discussion
During the initial stages of the design process, a significant amount of time was spent analyzing
the possible form factors of a GAVN-equipped craft. Several unique concepts were considered;
included among them were: a single-engine, gyroscopically balanced craft that would utilize a counterrotating disk to mitigate the effect of torque from the motor; a twin-engine design that resembled
hover boards from science fiction; and a four-engine craft that very closely resembled the quadcopters
that are now widely available to consumers. The first two designs were rejected due to concerns of
stability. While it would certainly be possible for them to maintain straight-and-level flight, the required

59

control and programming coupled with the relatively low power output made the one- and two-engine
designs impractical. The craft that used four GAVNs would be the most stable; however, the cost of
each individual engine would likely drive the total cost of the craft above the proposed budget.
In the end, it was decided that three GAVNs would be the most desirable. As is the case with
any craft that utilizes an odd number of engines, only two of the engines can contra-rotate, and
therefore an unbalanced torque will result from the one unmatched engine. To account for this, two
options were considered: one was angling the rear GAVN slightly, and the second was incorporating
contra-rotating propellers within each GAVN to ensure that each torque applied to the craft remains
balanced. Contra-rotating ducted fans were not available, so GAVN units will be angled during flight
to compensate (which will not be a significant issue given that they are designed to constantly change
angles normally).
Once the initial design had been decided, the team submitted an abstract to the 2013 AIAAYPSE conference, as discussed above. The abstract was accepted, and so on November 15 Chris,
Graham, Matthew, and Robert traveled to Laurel, Maryland to present the GAVN concept. While the
presentation did not receive any awards, the conference experience provided invaluable feedback from
professionals in the aerospace industry. It also forced the team to have the bulk of the core concepts
decided upon long before the Drexel Senior Design deadline.
At the conference, the team members interacted with aerospace professionals and were able
to receive constructive criticism on many key aspects of the project and concept. The team also got
to witness what other students in similar programs are working on, and where the aerospace industry
in general is headed. Feedback from judges who observed the teams presentation was particularly
helpful. One of the most significant comments addressed the need to obtain data on current
quadcopters to compare the GAVN prototype craft against. The teams presentation made several
references to how the GAVN will be better, in the specific areas of speed, maneuverability, and
stability, but the judges pointed out how measuring the abilities of a current product to facilitate
quantitative comparison would be advantageous. Outside of that feedback, the judges were very
impressed with the progress made over just eight weeks, specifically noting the quality of the dynamic
analysis and overall team effort.
At the conclusion of the winter term, the team was deep into the design process. All required
components that could not be manufactured by the team, such as the ducted fans, controllers, wiring,
and servos, had been selected and acquired. All detailed design for the GAVN unit, including FEA
analysis and using the 3D printer for verification and testing of the design, had been completed. Final
60

components had been printed on MEM department printers for use on the prototype aircraft. The
airframe was also completed through its first iteration of design, but required additional testing to
verify stability and structural integrity. Lastly, development of a control system for the prototype
aircraft was proceeding at an acceptable rate.
While the team did not reach all of the goals it set during fall term, a significant amount was
achieved during the spring term. The airframe was machined and verified to meet the teams
requirements for balance and accommodating all of the other design components. Code development
made remarkable progress, with the team overcoming significant hurdles in facilitating real-time
control of the aircraft over an ad-hoc wireless network. Towards the end of the term, multiple issues
with the inner-ring gears arose on certain GAVN units, but were successfully resolved in time for final
testing. The prototype aircraft was able to lift off and hover with only a minimal amount of external
support (for safety purposes only) as well as maneuver on a smooth floor, demonstrating its ability to
execute both rotational and translational movement while staying level.

Summary & Conclusions


From the research compiled, several issues with current VTOL technology were identified and
a possible solution was found which meets the criteria of the problem statement and addresses all of
the primary needs. A method for using direction controlled steering that is flexible with regard to its
propulsion method has been identified, and the feasibility of a prototype system and a prototype
aircraft using this method of steering has also been identified. By constructing a gimbal, placing a
thruster in its center, and by actuating the gimbal rings rotation with servomotors, a new method of
thrust vectoring that uses flexible engine types can be demonstrated, thereby adding capability to
VTOL aircraft which does not currently exist in a single scalable platform. While development did not
reach the complete goals set at the projects inception due to unexpected programming complexity,
the prototype aircraft did demonstrate that a GAVN-equipped aircraft was capable of taking off
vertically as well as maintain level flight during rotational and translational maneuvers, meeting four
of the five primary needs (with the unproven need being maintain absolute position against dynamic
external forces, which is theoretically possible but remains unproven).

61

Future Work
The future work of this project is divided into two main sections: work that can be done next
year by another senior design team and work that could be done further into the future. After the
completion of this project, there were some elements of the design that can be improved upon in
future iteration, including but not limited to gear design, gear materials, and batteries used. This work
is something that can plausibly be done by a senior design team in the coming year. This project also
requires future work developing a stable control algorithm to allow for the aircraft to autonomously
stabilize itself, something that could not be completed this year due to a lack of programming
knowledge among the team members. A future senior design team should be capable of attaining this
goal through the help of a computer science student or another student who is more capable with
programming.
The GAVN design has additional potential that could be developed in a few years time. One
idea involves creating a new version of the GAVN system that utilizes an alternate thrust source such
as a model size jet turbine engine, further proving that the GAVN system is thrust source independent.
Another future project is the development of a self-contained universal GAVN (SCU-GAVN) unit.
This is a design that could be attached to objects of various sizes and weights in multiple places, and
would be capable of creating an ad-hoc networked system of SCU-GAVNs that would be able to lift
and move the object with no preprogrammed data about its characteristics. While this idea is a number
of years off, it is believed to have significant potential and should be explored further.

Acknowledgements
The team would like to recognize the support of the following people and organizations for
their contributions to the GAVN project:

Dr. Ajmal Yousuff, for his academic and financial support, without which this project
would not have been possible.

The Boeing Company, for their generous financial support of the project.

National Instruments, for loaning the team the myRIO controller.

Brandon Terranova, for his assistance and expertise in producing all the 3-D printed
parts used in the GAVN construction.

The Drexel Machine Shop, for the use of their CNC machine and general guidance.

62

The coordinators and staff of the American Institute of Aeronautics and Aerospace
for allowing the team to present the GAVN at the YPSE conference in Virginia.

References
[1] NASA, 1995, "Shuttle Main Engine Firing in Gimbal Test," S. M. E. F. i. G. T.-. GPN-2000000552.jpg, ed., pp. A close-up view of a Space Shuttle Main Engine during a test at the John C.
Stennis Space Center shows how the engine is gimballed, or rotated to evaluate the performance of
its components under simulated flight conditions.
[2] Martin, L., 2013, "F-35B Short Takeoff/Vertical Landing Variant,"
http://www.lockheedmartin.com/us/products/f35/f-35b-stovl-variant.html.
[3] Hambling, D., 2012, "Armed Quadrotors Are Coming," Popular Mechanics.
[4] Britt, 2012, "How Much Does it Cost to Build and Fly a Quadcopter?," OddCopter.
[5] Brown, H. A. S. I. a. H., "Control of Jet Engines," pp. 1043-1059.
[6] Wikipedia, 2013, "Energy Density."
[7] NASA, "Dr. Robert H. Goddard, American Rocketry Pioneer,"
http://www.nasa.gov/centers/goddard/about/history/dr_goddard.html#.Upo5ZsR03To.
[8] Norden, L. O., 2006, Harrier II, Validating V/STOL, Naval Institute, Annapolis, Maryland.
[9] Cooper, R. G., 1993, Winning at New Products: Accelerating the Process from Idea to Launch,
Basic Books, Cambridge, Massachusetts.
[10] Robert G. Cooper, S. J. E. a. E. J. K., 2007, "Optimizing the Stage-Gate Process."
[11] Mango, N., 2004, "Measurement and Calculation of Vehicle Center of Gravity Using Portable
Wheel Scales"," SAE 2004 World Congress & Exhibition, SAE.

63

Appendices
Appendix A: Assorted Images
GAVN
Drexel University

Today's Date:

5/22/2014 Thursday
(vertical red line)

Project Lead: Matthew Brenner


Start Date: 9/1/2013
Sunday

9/23/13

12/2/13

71

100%

51

71

Proj. Scope

9/24/13

9/29/13

100%

Deliv erables Def.

9/23/13

9/26/13

100%

Clarify Focus

2.3.1
2.3.2

9/23/13

9/29/13

100%

2.4

Preproposal Submission

9/24/13

9/30/13

100%

Phase 2 - Arch & Planning

10/1/13

12/2/13

63

100%

45

63

3.1

Concept Dev .

10/1/13

11/9/13

40

100%

29

40

10/1/13

11/9/13

40

100%

29

40

10/1/13 10/30/13

30

100%

22

30

3.1.1

Hist. Tech. Report

3.1.1.1

Research

3.1.1.2

Compilation of Research

3.1.1.3

Write Report

3.1.2
3.2
3.2.1

Stakeholders and Needs


Preliminary Design Dev .
Basic Design Decisions

10/30/13

11/3/13

100%

11/3/13

11/9/13

100%

10/1/13

10/4/13

100%

10/20/13 10/26/13

100%

10/20/13 10/26/13

100%

3.2.1.1

# Nacelles

10/20/13 10/26/13

100%

3.2.1.2

Shape

10/20/13 10/26/13

100%

11/25/13

12/2/13

100%

3.3

Proposal Report

3.3.1

Report First Draft

11/16/13 11/29/13

14

100%

10

14

3.3.2

Report Final Draft

11/30/13

12/2/13

100%

Proposal Presentation

11/25/13

12/2/13

100%

3.4
3.4.1

Created Presentation

11/16/13 11/29/13

14

100%

10

14

3.4.2

Practice

11/30/13

12/2/13

100%

3.4.3

Present

12/2/13

12/2/13

100%

1/6/14

4
4.1
4.1.1
4.1.2
4.2

Phase 3 - Design & Dev

4/7/14

92

100%

66

92

10/20/13 11/15/13

27

100%

20

27

Mathmatical Model

10/20/13 11/15/13

27

100%

20

27

Control Report

10/20/13 11/15/13

Preliminary Control Theory

Main Winter Term Tasks

27

100%

20

27

1/6/14

3/14/14

68

100%

50

68

4.2.1

Schedule and assign tasks

1/6/14

1/12/14

100%

4.2.2

Choose Parts

1/9/14

1/19/14

11

100%

11

4.2.3

Order/Reciev e Parts

1/14/14

2/7/14

25

100%

19

25

4.2.3.1

Order Parts

1/14/14

1/20/14

100%

4.2.3.2

Receiv e All Parts

1/14/14

2/7/14

25

100%

19

25

4.2.4

Component Testing

2/7/14

2/20/14

14

100%

10

14

4.2.5

Mk 1 Gimbal Design

1/14/14

2/22/14

40

100%

29

40

4.2.6

Mk 1 Aircraft Design

2/23/14

3/8/14

14

100%

10

14

4.2.7

Elev ator Speech Script

1/14/14

3/3/14

49

100%

35

49

Winter Report

2/28/14

3/14/14

15

100%

11

15

3/31/14

4/7/14

100%

4.2.8
4.3

Main Spring Term Tasks

4.3.1

Schedule and assign tasks

3/31/14

4/6/14

100%

4.3.2

Mk 1 Assembly

3/17/14

3/31/14

15

100%

11

15

4.3.3

Mk 1 Wiring Diagram

3/31/14

4/7/14

100%

Phase 4 - Test & Opt.

3/31/14

5/19/14

50

92%

36

45

5.1

Testing

3/31/14

5/19/14

50

100%

36

50

Bench Testing

3/31/14

4/13/14

14

100%

10

14

5.1.1.1

Motor Tests

3/31/14

4/13/14

14

100%

10

14

5.1.1.2

Sensor Tests

3/31/14

4/13/14

14

100%

10

14

4/7/14

5/19/14

43

100%

31

43

5.1.1

5.1.2

Flight Testing

5.1.2.1

Stable Flight

4/7/14

5/19/14

43

50%

31

21

22

5.1.2.2

2D Manuv ering

4/7/14

5/19/14

43

100%

31

43

5.1.2.3

Hov er Testing

4/7/14

5/19/14

43

100%

31

43

Phase 5 - Launch

5/19/14

5/19/14

0%

6.1

Future Work

5/19/14

5/19/14

0%

AIAA YPSE Conference

10/20/13 11/15/13

27

100%

20

27

7.1

Submitt Project Abstract

10/20/13 10/22/13

100%

7.2

Prepare Presentation

11/1/13 11/14/13

14

100%

10

14

11/1/13 11/10/13

10

100%

10

11/10/13 11/14/13

100%

11/15/13 11/15/13

100%

7.2.1
7.2.2
7.3

Created Presentation
Practice
Present at AIAA Conference

19 - May - 14

Find Sponsors

2.3

12 - May - 14

2.2

28 - Apr - 14

15

05 - May - 14

11

21 - Apr - 14

100%

14 - Apr - 14

15

07 - Apr - 14

10/7/13

31 - Mar - 14

9/23/13

24 - Mar - 14

GANTT chart

17 - Mar - 14

2.1

10 - Mar - 14

71

03 - Mar - 14

10

51

24 - Feb - 14

100%

17 - Feb - 14

100%

71

10 - Feb - 14

10

12/2/13

27 - Jan - 14

9/23/13

9/23/13

03 - Feb - 14

9/14/13

Phase 1 - Scoping

20 - Jan - 14

Form Team

13 - Jan - 14

1.3

06 - Jan - 14

14

30 - Dec - 13

10

23 - Dec - 13

100%

16 - Dec - 13

14

09 - Dec - 13

14

9/14/13

02 - Dec - 13

10

9/1/13

25 - Nov - 13

Days Remaining

23

100%

Find Adv isor

18 - Nov - 13

Days Complete

16

14

1.2

11 - Nov - 13

Working Days

100%

9/14/13

28 - Oct - 13

% Complete

23

9/1/13

Start

04 - Nov - 13

Duration (Days)

9/23/13

Init. Project Formation

21 - Oct - 13

End

9/1/13

1.1

Task
Lead

14 - Oct - 13

Tasks
Phase 0 - Discovery

07 - Oct - 13

WBS
1

30 - Sep - 13

First Day of Week (Mon=2):

[42]

Figure 47: Gantt Chart

64

Figure 48: GAVN Comparative Thrust Weight Energy Analysis

65

Figure 49: Quadcopter Dynamics and Control LabVIEW Example VI

66

Figure 50: Quadcopter Dynamics and Control Subsystem VI Integrations

Figure 51: EDF Model Rendering

67

Figure 52: Servomotor Model Rendering

Figure 53: Servomotor Photograph

Figure 54: Inner Gimbal Ring FEA - Stress due to Max Thrust

68

Figure 55: GAVN Mark 1 Design

69

Appendix B: Thrust Test Data


Test 1
2/2/2014
1910
RMS
Thrust
Full
Thrust
(V)
(g)
(V)
(N)
1.986
58
2.81
0.57
2.258
78
3.19
0.77
2.547
100
3.60
0.98
2.798
120
3.96
1.18
2.819
146
3.99
1.43
3.461
188
4.89
1.84
3.645
208
5.15
2.04
3.797
228
5.37
2.24
3.982
250
5.63
2.45
6.14
288
8.68
2.83
Test 3
2/2/2014
1915
RMS
Thrust
Full
Thrust
(V)
(g)
(V)
(N)
1.547
38
2.19
0.37
1.661
42
2.35
0.41
Test 4
2/14/2014
1920
RMS
Thrust
Full
Thrust
(V)
(g)
(V)
(N)
1.957
58
2.77
0.57
2.797
122
3.96
1.20
3.051
146
4.31
1.43
3.247
168
4.59
1.65
3.495
190
4.94
1.86
3.827
226
5.41
2.22
5.87
248
8.30
2.43
6.1
292
8.63
2.86
6.38
304
9.02
2.98
6.55
352
9.26
3.45
6.65
374
9.40
3.67

Test 5
2/3/2014
RMS
Thrust
(V)
(g)
1.581
36
1.963
56
2.197
70
2.495
90
2.766
112
3.043
136
3.224
154
3.399
174
3.622
196
3.759
212
3.902
238
6.18
268

Full
(V)
2.24
2.78
3.11
3.53
3.91
4.30
4.56
4.81
5.12
5.32
5.52
8.74

6.48
6.72
6.92
7.09
7.22

298
344
368
384
426

9.16
9.50
9.79
10.03
10.21

7.53
7.56
8.23
8.52
9.08

458
462
568
608
700

10.65
10.69
11.64
12.05
12.84

Test 7
1825
Thrust
(N)
0.35
0.55
0.69
0.88
1.10
1.33
1.51
1.71
1.92
2.08
2.33
2.63

2/3/2014
RMS
Thrust
(V)
(g)
1.902
54
2.128
66
2.364
84
2.635
104
2.873
122
3.168
144
3.324
164
3.538
184
3.669
192
3.812
214
5.83
238
5.99
250

1840
Thrust
(N)
0.53
0.65
0.82
1.02
1.20
1.41
1.61
1.81
1.88
2.10
2.33
2.45

Full
(V)
2.69
3.01
3.34
3.73
4.06
4.48
4.70
5.00
5.19
5.39
8.24
8.47

2.92
3.37
3.61
3.77
4.18

6.15
6.45
6.62
6.84
7.09

270
284
324
360
396

8.70
9.12
9.36
9.67
10.03

2.65
2.79
3.18
3.53
3.88

4.49
4.53
5.57
5.96
6.87

7.42
7.54
8.1
8.3
8.43
8.78
9.02

452
468
536
566
602
676
708

10.49
10.66
11.46
11.74
11.92
12.42
12.76

4.43
4.59
5.26
5.55
5.91
6.63
6.95

70

Test 8
2/3/2014
RMS
Thrust
(V)
(g)
3.257
162
3.482
180
3.787
212
5.99
246
6.07
264
6.46
292
6.65
344
6.76
358
7
396
7.1
410
7.22
432
7.31
446
7.41
462
7.95
552
8.19
562
8.24
592
8.34
598
8.72
672
8.92
706

Full
(V)
4.61
4.92
5.36
8.47
8.58
9.14
9.40
9.56
9.90
10.04
10.21
10.34
10.48
11.24
11.58
11.65
11.79
12.33
12.61

Test 10
1848
Thrust
(N)
1.59
1.77
2.08
2.41
2.59
2.86
3.37
3.51
3.88
4.02
4.24
4.38
4.53
5.42
5.51
5.81
5.87
6.59
6.93

2/3/2014
RMS
Thrust
(V)
(g)
2.507
96
2.773
118
2.979
134
3.198
156
3.381
174
3.522
194
3.708
208
3.962
238
6.01
254
6.27
272
6.33
280
6.39
288
6.5
332
6.62
346
6.86
382
7.06
412
7.19
426
7.28
446
7.72
512
7.93
534
8.01
550
8.17
578
8.25
598
8.48
622
8.76
678
8.81
704
8.9
722

Full
(V)
3.55
3.92
4.21
4.52
4.78
4.98
5.24
5.60
8.50
8.87
8.95
9.04
9.19
9.36
9.70
9.98
10.17
10.30
10.92
11.21
11.33
11.55
11.67
11.99
12.39
12.46
12.59

1900
Thrust
(N)
0.94
1.16
1.31
1.53
1.71
1.90
2.04
2.33
2.49
2.67
2.75
2.83
3.26
3.39
3.75
4.04
4.18
4.38
5.02
5.24
5.40
5.67
5.87
6.10
6.65
6.91
7.08

71

Appendix C: Tricopter Flight Dynamics Code


%% Tricopter Flight Dynamics
%% Dimensions
L= 2;
% side length of equil. triangle
h= 0.5*sqrt(3)*L;
% height of equil. triangle
r1= [2*h/3 0 0]';
% vectors from each GAVN to body CG
r2= [-h/3 L/2 0]';
r3= [-h/3 -L/2 0]';
%% Thrust Vector Transformation of Each Motor
%%%%%%%% change to Y-shape, need yaw rotation for front GAVNS %%%%%%%%%%%%%
Rx1= [1 0 0; 0 cos(phi1) sin(phi1); 0 -sin(phi1) cos(phi1)];
%roll
Ry1= [cos(thet1) 0 -sin(thet1); 0 1 0; sin(thet1) 0 cos(thet1)];
%pitch
R1= (Ry1*Rx1);
%body frame to GAVN frame
Fb1= R1\[0 0 -T1]';
Rx2= [1 0 0; 0 cos(phi2) sin(phi2); 0 -sin(phi2) cos(phi2)];
%roll
Ry2= [cos(thet2) 0 -sin(thet2); 0 1 0; sin(thet2) 0 cos(thet2)];
%pitch
R2= (Ry2*Rx2);
%body frame to GAVN frame
Fb2= R2\[0 0 -T2]';
Rx3= [1 0 0; 0 cos(phi3) sin(phi3); 0 -sin(phi3) cos(phi3)];
%roll
Ry3= [cos(thet3) 0 -sin(thet3); 0 1 0; sin(thet3) 0 cos(thet3)];
%pitch
R3= (Ry3*Rx3);
%body frame to GAVN frame
Fb3= R3\[0 0 -T3]';
%% Forces on Body
m= 1;
% mass of body (1 as placeholder)
g= 9.81;
% gravity
% Cx= [1 0 0; 0 cos(roll) sin(roll); 0 -sin(roll) cos(roll)];
%roll
% Cy= [cos(pitch) 0 -sin(pitch); 0 1 0; sin(pitch) 0 cos(pitch)];
%pitch
% Cz= [cos(yaw) sin(yaw) 0; -sin(yaw) cos(yaw) 0; 0 0 1];
%yaw
Fgrav= m*g*[-sin(pitch) cos(pitch)*sin(roll) cos(roll)*sin(pitch)]';
Fb= Fb1 + Fb2 + Fb3 + Fgrav;
% forces in body frame
%%%%% separate gravity from F and include explicity in accel eqns %%%%%%%%%
%% Moments on Body
%%%%%%%%%%%%%%%%%%%%%%%%%
%%% needs work %%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%
% moment arms(CG to GAVN)
l1= -r1; l2= -r2; l3= -r3;
% moment due to thrust
Mth= cross(l1,Fb1) + cross(l2,Fb2) + cross(l3,Fb3);
%% Inertia
% placeholder identity matrices for GAVN inertias
Ib= eye(3);
% inertia of body
Ig1= eye(3);
% GAVN#1 inertia
Iv1= R3*Ig1/R1;
% rotate to body frame
Ig2= eye(3);
% GAVN#2 inertia
Iv2= R3*Ig2/R2;
Ig3= eye(3);
% GAVN#3 inertia
Iv3= R3*Ig3/R3;
% Parallel Axis translation
m1= 1;
% mass of each GAVN unit (placeholder)
m2= 1;
m3= 1;
Icg1= Iv1 + m1*(norm(r1)*eye(3) - r1*r1');
Icg2= Iv2 + m2*(norm(r2)*eye(3) - r2*r2');
Icg3= Iv3 + m3*(norm(r3)*eye(3) - r3*r3');
J= Ib + Icg1 + Icg2 + Icg3; % total inertia about CG

72

%% Kinematics
% Translational Motion
%%%%%%%%%% include gravity explcitly, separate from F %%%%%%%%%%%%%%%%%%%%%
udot= Fb(1,:)/m + [r*v - q*w];
vdot= Fb(2,:)/m + [p*w - r*u];
wdot= Fb(3,:)/m + [q*u - p*v];
% Rotational Motion
omega= [p; q; r];
hb= J*omega;
% body angular momentrum
% GAVN angular momentums, rotated to body frame
hg1= [0; 0; Jr1*om1];
hg1b= R1\hg1;
hg2= [0; 0; Jr2*om2];
hg2b= R2\hg2;
hg3= [0; 0; Jr3*om3];
hg3b= R3\hg3;
%%%%%%%%%%% translate rotor ang momentum to body CG %%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%% include in omegadot equation %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%% differentiate J in labview %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
omegadot= J\(M - Jdot*omega - cross(omega,hb));
Pdot= omegadot(1,:);
Qdot= omegadot(2,:);
Rdot= omegadot(3,:);

73

Você também pode gostar