Development of a Low Cost 3-Axis Reaction Wheel System

for Ultra-Fine Satellite Pointing Capabilities

Final Technical Report

The George Washington University
Department of Mechanical and Aerospace Engineering

Abigail DeMasi
Jordan Giacoma
Joseph Joyce
Cesar Laurent
John Stricker
 Table of Contents

Executive Summary 2

I - Introduction 3
Scope of the project 3
Review of technical literature 4
Fundamental principle 4
Reliability 4
Temperature range 5
Dead-zone 5
Testing Device 5
Design requirements 5

II - Design Description 6
Summary of Design 6
Detailed Description 9
Motors & Controller 9
Housing 12
c) Zero Resistance Testing Device Error! Bookmark not defined.

III - Evaluation & Testing 19

Summary of evaluation and testing 19
Subsystem testing 20
Motors & Controller 20
Housing 20
Zero Gravity Testing Device Error! Bookmark not defined.
3) Full system testing Error! Bookmark not defined.

IV - Summary and Recommendations 23

Summary 23
Recommendations 24

V - References 25

VI - Appendices 27
Appendix A - Momentum requirement for the x,y and z axis 27
Z - axis 27
X and Y - axis 27

1
Abigail DeMasi: Executive Summary, Scope of the Project, Fundamental Principle, Summary of Design,
Summary of Evaluation and Testing
Jordan Giacoma: Scope of project, review of technical literature, Design Requirements (introduction),
Detailed description of Design, Detailed description (Evaluation & Testing)
Joseph Joyce: Housing Design Description, Housing Evaluation & Testing
Cesar Laurent: Formatting ,Executive Summary, Introduction, Motors & Controller
John Stricker: Testing, Recommendations, Design Description
Finalists at the GW School of Engineering and Applied Science R&D Showcase

Executive Summary
The development of CubeSat nanosatellites has made space data collection significantly more
accessible to the public with its inexpensive modular design, however improvements in further decreasing
costs still need to be done in order to sufficiently reduce the economic barriers associated with space
research. Many of the subsystems needed for space research missions are still financially out-of-reach for
some. For example, Reaction Wheel Systems (RWS) are an integral component to most satellite Attitude
Determination and Control Systems (ADCS). They provide active control capabilities by improving
pointing accuracy, enhancing satellite stability and promoting more precise orbital station keeping.
Therefore, the Low Cost 3-Axis Reaction Wheel System for Ultra-Fine Satellite Pointing Capabilities aims
to create a more economical reaction wheel system for satellites compared to what is currently available.
Reaction Wheel Systems are composed of flywheels that each spin at different speeds; when rotation speed
is changed, a spacecraft counter-rotation proportionate and opposite to the rotation of the flywheel is
induced through the conservation of angular momentum. In order to be applicable future CubeSats, the
system must meet the following requirements:

- Must fit into PC/104 dimensions (90x96mm).

- Microcontroller integration.
- Sensors to measure and adjust for changes in rotational speed.
- Must be able to change angular velocity from 2°/s to 0.1°/sec (or less).
- Motors must be able to reverse.
- Overall weight of the system should be less than 150g.
- Overall budget of the system should be less than $500.

2
 - The system will be integrated at ± 10 cm from the center of mass on the x and y axes.
- The system will be integrated at ± 5 cm from the center of mass on the z axes.

The Reaction Wheel System is composed of three (3) Faulhaber 1509B motors and three (3)
SC1801S Electronic Speed Controllers or ESC’s (one for each axis). The components were integrated on a
3D-printed housing made from polylactic acid (PLA) and connected to a microcontroller unit or MCU for
testing. The Reaction Wheel System was tested using a 3D-printed Acrylonitrile Butadiene Styrene (ABS)
Zero Resistance testing device specifically designed for the purpose of studying the torque produced by the
system. The testing results were successful in proving that the system is capable of producing the torque
necessary to lower the angular velocity of the CubeSat on every axis by the required value.

I - Introduction

1) Scope of the project

CubeSats are a class of microsatellite primarily used for research purposes. Due to their simple,
modular design, consisting of one or more 10x10x10cm cubic units, they can be customized to meet the
needs of individual research projects. While this can be accomplished for a fraction the cost of traditional
satellites, many of the systems and components needed for mission success are still prohibitively expensive.
For example, a 3 axis reaction wheel system is essential for attitude determination and control. The system
allows for detumbling and ultra-fine pointing of the spacecraft; two abilities essential to many space
research missions. CubeSat-compatible reaction wheel systems can cost upward of $7,100 [1]. The
development of CubeSat technology has made space data collection significantly more accessible to the
public, although work on reducing the costs of essential system components must still be done. This will
alleviate much of the current financial strain and further promote space research.
The objective of this work is to fully understand the purpose and function of reaction wheel
systems. This will enable us and others to design more cost effective alternatives and consequently reduce
the economic barriers faced by many space research missions.
The system primarily consists of off-the-shelf components, a 3D printed housing, as well as the
electronics necessary to control the motors. It must be capable of rotating a 3U (or three unit) cube satellite
along any of its three axes by precise amounts. This is necessary because the plasma thrusters used for
primary flight control have a minimum thrust output that yields a rotation of 2∘ per second. This results in a
dead zone wherein the CubeSat can only oscillate back and forth. The reaction wheels slow this rotation
from the aforementioned 2∘ per second to 0.1∘ per second. The motors accomplish this by employing the
fundamental physical axiom of the conservation of angular momentum. The reaction wheel system will be
integrated into George Washington Universities CubeSat Program and eventually be installed in the 3U
CubeSat. Before delivering the final product to the CubeSat Program, the system will be tested in a Zero
Resistance Simulation Device (Z.R.S.D.) designed for small payloads under 150 grams. With a budget of

3
five hundred dollars ($500), the cost of the system is only a small percentage of the cheapest available
reaction wheel systems on the market.

2) Review of technical literature

Fundamental principle
A Reaction Wheels System (RWS) uses the fundamental principle of conservation of momentum
to control spacecraft attitude. This allows them to physically rotate, or point, the spacecraft by very small
amounts. Because of this, the reaction wheels must be capable of spinning fast enough to match the
spacecraft’s angular momentum and generate torques capable of rotating the spacecraft by very small
amounts [2]. In other words, they act as a sort of storage unit for the spacecraft’s angular momentum.
Reaction wheel systems generally consist of a small electric motor capable of continuous rotation
in both the forward and reverse directions, and a flywheel [3]. Both components are specifically engineered
to fulfill the spacecraft attitude control requirements [4]. Although, the purpose of the flywheel is to add
mass to the system, and therefore increase the system’s inertia; the design of the flywheel depends on the
performances of the motor. Thus, the first step of a RWS design process is to carefully select the motors to
fulfill requirements such as reliability, temperature range, dimensions, sensors, rotation speed, moment of
inertia and cost.
Most spacecraft are equipped with four individual reaction wheels positioned in a tetrahedron
configuration to allow for rotation about three distinct axes while still being tolerant to the failure of one
wheel [5]. The placement of the reaction wheels is a key component in enabling the spacecraft to rotate
about three unique axes and orient itself in a desired direction.
In order to point the spacecraft in one direction, both the speed of the reaction wheel corresponding
to the proper axis of rotation should be increased proportional to the desired amount of rotation, and the
direction of motor rotation should be opposite to the direction of the desired rotation [6]. It would follow
then, that in order to rotate the spacecraft in the other direction the speed of the reaction wheel should be
decreased proportional to the desired rotation.
It is possible for reaction wheels to reach their maximum rotational speed and become ‘saturated’
due to the addition of momentum as a result of external forces such as solar radiation, gravitational forces,
magnetic fields, and micrometeorites [7]. In this case, the system must be ‘desaturated’ meaning the excess
momentum is removed from the system by applying a large torque to the spacecraft [7].

Reliability
A typical CubeSat mission is estimated to last between three and twelve months. As a consequence,
the motors of the RWS must be able to operate continuously during a period of twelve months. Considering
this twelve month reliability standard along with the limited budget of five hundred dollars, it was decided
that the RWS should be designed using commercial off-the-shelf motors. In space applications, it is
important to minimize friction while maximizing efficiency and durability. Brushless Direct Current
(BLDC) motors match those selection criteria. BLDC do not have any brushes, meaning that the operating
lifetime of the motor is not limited by bearing failure and that the torque response is linear [8]. Moreover,
BLDC motors have a higher efficiency compared to regular Alternating Current motors [9].

4
Temperature range
During their mission into space, CubeSats are exposed to an extreme range of temperatures while
orbiting around Earth. The outside temperature of the nanosatellites soar to 121°C when facing the sun and
plunge to -157°C when passing around the dark side of Earth [10] , and it is estimated that components
inside the CubeSat are exposed to temperature between -30°C and +85°C [11]. As a consequence, motors
must be approved for continuous operations within this temperature range.

Dead-zone
In CubeSats using a plasma thruster system [12] and a RWS for spacecraft attitude control, thrusters
are used to complete large changes in angular speed (more than 2°/sec) while the RWS is used to operate
small changes in angular speed (less than 2°/sec). In other words, by using only the plasma thrusters system,
the spacecraft attitude control cannot conduct operations at small angular speeds and the RWS needs to be
used to operate in the thrusters dead-zone, a range of angular speed that thrusters cannot maintain [13]. In
order to select the right motor for the RWS, one needs to know the Moment of Inertia I, as well as the
continuous operations Angular speed of our motors. Since Angular speed is a function of the Moment of
Inertia I and the Torque H, we first need to find the Torque of the CubeSat at an angular speed of 2°/s to
know the maximum Torque value that the RWS have to overcome.

Testing Device
A testing device was needed in order to determine the effective rotational stopping potential of
our system. Various methods of accomplishing this goal were considered and several articles and thesis
papers were collected and studied. Comparing the strengths of each approach led to the selection of an air
bearing for our design. This allowed for the closest approximations of zero resistance motion and
translation. By levitating the payload upon a cushion of air one can truly see how their reaction control
system would behave in low earth orbit. [16]

3) Design requirements
The design requirements established by the GW CubeSat team are summarized in Table I-1.

Table I-1: Design requirement for GW CubeSat Reaction Wheel System

Requirement
Must fit into PC/104 dimensions (90x96mm)

Microcontroller integration

5
 Sensors to measure and adjust for changes in rotational speed

Must be able to change angular velocity from 2°/s to 0.1°/s (or less)

Motors must be able to reverse

Overall weight (of all 3) < 150g

Budget < $500

RWS will be integrated at ± 10 cm from the center of mass on the x and y

axes.

II - Design Description

1) Summary of Design
The reaction wheel system pictured in Figure II-1 is comprised of three 6-volt brushless DC micro
motors as well as three electronic speed controllers (ESC’s). Each motor is fixed to one face of a small box,
measuring 1.3 x 1.3 x 1.2 inches, so as to align each motor axle with each corresponding axis of rotation.
Likewise, each of the three electronic speed controllers is fixed to each of the three remaining sides of the
cube and secured with 1/16 inch screws.
An electronic speed controller is an electronic circuit that regulates the speed of each motor by
varying the voltage supplied to it. Generally, electronic speed controllers are rated according to their
maximum peak output current. The model used here is rated for a maximum of 2 Amps. The addition of
electronic speed controllers to the final product design enables the motors to be controlled with a more
precise variation of motor speed, something critical to achieving a change angular velocity from 2°/s to
0.1°/s (or less).
Traditionally, a reaction wheel system is composed of four motors, capable of both forward and
reverse rotation, and four flywheels attached to each motor axle [5]. The purpose of the flywheel is to add
supplementary mass in order to increase the inertia of the system. However, after calculating the maximum
torque of the motors it was determined that they could match the momentum requirements of .002424
kg*m^2 * rad/sec, .002424 kg*m^2 * rad/sec, and 0.0002543 kg*m^2 * rad/sec for rotation in the X, Y,
and Z axis, respectively (see Appendix A). This is beneficial because it eliminated the need to integrate a

6
flywheel to provide additional mass. Furthermore, the design only includes three motors in order to avoid
redundancies, extra weight, and eliminate the budgetary strain of purchasing a fourth motor.
In practice, each motor would be connected to a single electronic speed controller, which, in turn,
is controlled by the satellite’s on-board attitude determination and control system. However, for testing
purposes, the reaction wheel system is be controlled by a small microcontroller unit (MCU). The Arduino
Uno was selected as the microcontroller because of its ubiquity and the plethora of helpful online resources
available.
In addition to the MCU, a device simulating weightlessness is used for testing the design of the
reaction wheel system. The goal of the device is to create an environment where the reaction wheel system
is able to compensate for changes in speed, direction, and angle in order to maintain position with respect
to some stationary point. The testing device, pictured in Figure II-3 was made using additive manufacturing
techniques and constructed from acrylonitrile butadiene styrene (ABS) plastic. It is comprised of four
sections; a base with two inlets for pressurized air, a body to evenly distribute the flow of air while also
correcting for irregularities in flow, a cap to hold the lid piece in place, a lid perforated with many tiny
pores to allow air flow, and a floating cup containing the reaction wheel system which will be suspended
in a stream of pressurized air.

Figure II-1 - The Low Cost 3-Axis Reaction Wheel System for Ultra-Fine Satellite Pointing Capabilities

7
 Figure II-2- Assembled RWS

Figure II-3 - Zero Resistance Testing Device

8
 2) Detailed Description

a) Motors & Controller

The Reaction Wheel System (RWS) uses three brushless flat DC micromotors series 1509B from
Faulhaber. These 4-pole 6V motors are equipped with three Hall sensors for digital feedback. The RWS
contains a motor and an Electronic Speed Controller (ESC) SC1801 for each of the three axes. All three
motor-ESC pairs are connected to an Arduino Mega 2560 for power and control. As pictured in Figure II-
1, the Series 1509B motor has eight wires. Connections 1 to 3 are for controlling the motor’s three phases,
connections 4 to 5 are for the voltage supply (Ground and 5V) and connections 6 to 8 are for the Hall
sensors. The eight connections of the Series 1509B are plugged to respective connections on the screw
terminal block on the motor side of the SC 1801 Electronic Speed Controller, as illustrated in Figure II-2.

Figure II-4 - Series 1509B Connection function [14]

Figure II-5 - SC 1801 Connection function [15]

The Electronic Speed Controller provides feedback control between the Arduino signal and the
Hall sensors data. The accelerometer sends a signal to the Arduino, the Arduino converts the signal for the
Electronic Speed Controller, and the Electronic Speed Controller provides a signal to the motor that varies
according to the Hall sensors data. In other words, the desired speed sent by the Arduino signal should
match the actual speed received by the Electronic Speed Controller from the Hall sensors. The Arduino
Mega 2560 is connected to the screw terminal of the Electronics Speed Controller by five different
connectors as shown in Figure 3.

9
 Figure II-6 - SC 1801 Connection on the supply side [15]

The electronic supply pin (Up) provides a voltage for the feedback loop, while the motor power supply pin
(Umot) provides the voltage required to make the motor spin. Both connectors are connected to the 5V pin
of the Arduino through a breadboard. The control voltage for the set speed is provided by the Unsoll pin and
the switching input for the rotation direction of the motor is provided by the DIR pin. As opposed to Upand
Umot, which receive a constant signal, the Unsolland DIR pins receive a continuously varying signal. In other
words, the Unsoll pin receives a signal from the Arduino that is proportional to the desired motor speed, and
the signal that the DIR pin receives determine the rotation direction input; a voltage less than 0.5V enables
an anti-clockwise rotation while a voltage higher than 3V enables a clockwise rotation.
In order to test the Reaction Wheel System on one axis, an accelerometer connected to the Arduino
measures the rate of change in angular velocity of the particular axis at a current moment in time through
the Gyroscope function of the Arduino (Figure II-3). When the accelerometer is moving, the Arduino sends
a signal to the Electronic Speed Controller which provides a voltage to the motor that is proportional to the
angular velocity of the accelerometer and its direction of rotation. When the accelerometer is moving in the
opposite direction, the motor spins in the other directions.
The RWS controller was developed in Arduino using the Arduino PID library. This library contains
a robust PID equation with simply defined inputs and outputs, making it easy to use and modify for a variety
of applications. Though Simulink offers a wider range of tools and capabilities for controller development,
it is very expensive to purchase for commercial use and lacks support for common sensors and modules,
including the GY-521 accelerometer used in this system. By contrast, Arduino is a largely open-source
platform with broad hardware support. This makes it an ideal choice for this project, in which cost is a
primary concern.
The controller for the RWS accepts input data from attitude sensors and compares that data to a
desired reference value. While aboard the CubeSat, a star tracker will be used for attitude sensing; however,
for testing purposes a GY-521 accelerometer/gyroscope chip was used. The GY-521 is capable of precisely
measuring deflection on the x-, y-, and z-axes, making it ideal for the testing of this three-axis RWS. The
measurements are compared to a reference value of zero, which corresponds with zero degrees of deflection
from the setpoint. The controller then outputs an 8-bit integer that sets the duty cycle of pulse width
modulation (PWM) signal sent to the motor through the speed controller, allowing the motor speed to be
adjusted based on the angular acceleration detected by the gyroscope. This process is outlined in Figure 1.

10
 Figure II-7- Block diagram of testing configuration

The transfer functions of the motor, speed controller, and power source are too complicated and contain
too many unknowns to effectively create a mathematical model of by hand. Because of this, the values for
gains KP, KI, and KD may be determined through extensive trial and error. Based on the performance
requirements of this RWS, the optimal values of KP, KD, and KI are on the order of 10, 10-1, and 10-1,
respectively. It should be noted that the optimal gain values may change if the inertia of the system
containing the RWS changes. Further tuning of the gains will likely be necessary to achieve the desired
performance from the RWS depending on the mass of the spacecraft containing it.
Figure 2 shows the temporary test setup used during the controller design process. Because the torque
generated by each motor in the RWS is isolated on a single axis, each axis may be considered independently
of one another, and the controller for each axis has no effect on the function of any other axis. As a result,
each axis of the RWS can be tested individually; only a single motor is needed for the test setup. This also
greatly simplifies the controller design of the system: Each axis is identical aside from its total inertia, so
controllers for each individual axis can be developed independently and combined later.

Figure II-8- Photo of testing configuration

11
 b) Housing

Figure II-9- CAD Assembly of Housing (measurements in inches)

The housing, shown assembled in Figure II-9, was designed to be able to have the whole system to
be put together compactly and not have the motors and ESCs have one floating and be unprotected. It was
designed so the whole system would take up less than the area of a PC104 computer standard (90x96mm).
All three motors and ESCs are a part of the housing and it all is smaller than a PC104.
The housing was designed so that the motors and ESCs would not only fit onto the faces of the
housing, but also they could be attached by screws. The motors and the ESCs already had 2 mm screw holes
designed into them. The housing then had to be designed with precise and accurate measurements to
accommodate 2 mm screws.
To accommodate this need, the housing was manufactured by 3D printing. This made it easier to
manufacture with the precise and accurate measurements needed. It also decreased manufacturing time as
it was all automated and the CADs for it were already made. Additionally, the PLA filament used met
requirements as it is lightweight and would be able to withstand the stresses it would undergo when in use.

12
 Figure. II-10- One Part of Two Part Housing

As opposed to one simple box that was designed, instead two identically sized U shaped objects
were designed to be the housing as seen in Figure II-10. This was done so the motors can inserted from the
inside to reduce space taken up by the system. They would be slid together and form the housing. Even
after being assembled, the housing can be taken apart into two so modifications or maintenance can be
made. Additionally there are holes one every face. They were intended for the motors however they are still
on the sides with ESCs. Not only do they decrease weight, but the ESCs can be removed and it can be seen
if the motors are rotating inside by looking through the holes.

Figure II-11- Two-piece housing

c) Zero Resistance Testing Device (Z.R.T.D.)

A testing device was created in order to determine with a high degree of accuracy the revolutions
per minute necessary to achieve controlled and precise torques. The device allowed for full rotation
around the Z axis and roughly 25∘ of rotation about the X and Y. Several literary sources were consulted
throughout the design process and the major component of our device is based upon “pucks” that are
shown in the figure below:

13
 Figure II-12- Simulation Device [16]

The rotation was facilitated through a levitating surface that sat atop a cushion of air that was
streamlined and dispersed evenly in order to eliminate irregularities in the flow. The general dimensions
of the device shown in Figure II-12 are as follows:
● 5” in diameter.
● 8” in height.
● The central shaft starts with a 4” diameter and diminishes to 1” before returning to 4”.
● The walls of the device are at minimum 0.5” and maximum 1”.

Figure II-13- CAD assembly of Z.R.T.D.

In order to achieve the levitation necessary to imitate a zero gravity environment compressed air
was injected into the base. The turbulent air was collected and streamlined through the body after having

14
passed through the base. The body is structured in a way that the compressed air is further pressurized in
order to reduce as many irregularities in the flow as possible. This is crucial in that a turbulent,
unpredictable flow would result in what is called “shearing”. This is the phenomena of the levitating cup
temporarily touching the cap and shaking, further destabilizing itself. This can make the imitation of zero
resistance motion very inaccurate.
The cap was designed to facilitate levitation in the cup. By dispersing the air as evenly as
possible, a singular uniform cushion was created to act upon the section of the cup that was currently
above the cap. The cup held the three motors, the power supply, the Arduino, and the Bluetooth
transmitter. The various individual components of the testing device were held together through a locking
mechanism. One major reason for this development was the increase in customizability and ease. Coming
up with a uniform system that could be easily ported between CAD files allowed for quick modifications
to be made when finalizing the design. This was especially useful when varying the diameter, number,
and spacing of the holes in the cap. These alterations modified the size of the levitation cushion and
changed the rotation characteristics of the floating cap.
In summation, all aspects of the assembly were planned and fitted to ensure maximum
customizability and accuracy in creating a zero resistance environment. Further, each aspect was able to
be reprinted and adjusted with relative ease and aided in reaching our delivery time. The following figures
illustrate the individual components in detail:

Figure II-14-Base

Figure II-15-Body

15
 Figure II-16-Cap

Figure II-17-Lid

Figure II-18-Pressure Tap

16
Figure II-19-Compressed Air Release

Figure II-19-Testing Picture A

17
Figure II-20-Testing Picture B

Figure II-21-Testing Picture C

18
 Figure II-22-Testing Picture D

Figure II-23-150 Gram Weight

III - Evaluation & Testing

1) Summary of evaluation and testing

The RWS was tested using a setup designed to simulate the weightlessness of low earth orbit as
well as to play the role of the satellites’ onboard attitude determination and control system. The device used
to simulate weightlessness, shown in Figure II-12, was manufactured out of ABS plastic using additive
manufacturing techniques. The goal of the device was to create an environment where the effects of the
torques created by the motors would be visible. This was done by manufacturing the device in multiple
parts. First, a base with two inlets for injecting pressurized air into the device, a body piece to evenly
distribute the flow of air, a cap to hold a lid piece, a lid with perforated with several small holes, and finally
a floating cup containing the reaction wheel system. During the testing phase, the testing device performed
as planned and enabled the effects of the torques generated by the motors to be analyzed.

19
 The testing setup, seen in Figure II-12, additionally included a microcontroller unit as well as a 3
axis accelerometer-gyroscope module. The microcontroller was used to the run the PID controller and
power the motors; the final version of the device will be powered and controlled by the CubeSAT’s main
power bus and onboard computer, so the microcontroller will not be needed.
Testing determined that the RWS was capable of detecting rotation and responding to it by
activating the motors. Due to the relatively low angular momentum requirements for this RWS, it was
initially difficult to demonstrate if the motors were actually reducing the system’s rotational speed;
however, the addition of flywheels to the motors increased their capacity to counteract rotations of the
system significantly.
Further testing with heavier flywheels will likely be necessary to provide more conclusive test
results and allow for better tuning of the PID controller. As it was, the motors had to spin at full speed to
produce any noticeable effect on the rotation of the system. With more weight added to the flywheels, the
system will respond more to changes in motor speed, making it easier to tell if it is functioning correctly.

2) Subsystem testing

a) Motors & Controller

Basic testing of the motors and speed controllers in conjunction with the PID controller was
accomplished using the test setup shown in Fig. II-5. This configuration was primarily used to ensure that
the gyroscope was accurately reading angles, and to test the PID controller for instability. To test the
gyroscope, the breadboard holding the gyroscope was tilted, and the output was examined. In particular, it
had to be verified that the controller would spin the motors in the correct direction after receiving data from
the gyroscope. Once this was verified, the controller was ready for integration into the complete system.

b) Housing

20
 Figure. III-1- First Iteration of Housing

Many iterations of the housing were done. The original design was a simple 1 in. cube with holes
and screw holes on every side of the cube for the motor shown in Figure III-1. However, it was realized
more space could be saved if the motors were put inside the box instead of sitting on the outside of it. A
second iteration was designed as two U shaped piece that could be slid together like the final iteration.
The two pieces had to be designed to fit together and also be the same size and shape as the first
iteration. The side of the cube were chamfered by 45 degrees each so they would fit together in the same
size as the original iteration.
After it was printed, it was realized that although the pieces of the iteration fit together, it was
actually a rectangular prism as a opposed to a cube as it was longer in one axis. This was a result as the
sides were longer as the thickness of the faces were not taken into consideration.
The sides were then shortened in the third iteration. Additionally, ABS filament was used instead
of PLA as it is lighter. This iteration was printed and assembling the whole system was attempted.
However a downside to ABS is that is much more brittle that PLA. This was found as when the
system was being assembled, a piece of the housing broke attempting to put screws into it. Additionally it
was decided then that the ESCs would fit onto the housing. The housing, which was still a 1 in. cube, was
too small for the ESCs as they were larger and would overhang from the faces.
A final iteration was designed that was made out of PLA as opposed to ABS used in the the previous
iteration. The weight gain from going back to PLA was minimal especially in comparison to the mass of
the whole system as the housing is much lighter than the motors. Additionally the box was lengthened in
two axis to accommodate an ESC on each axis making it a rectangular prism as opposed to a cube. It was
still significantly smaller than a PC104. Additionally screw holes for a motor were replaced on three faces
and instead there were ones for the ESC. This iteration was printed and the whole system was assembled
into it and screwed in.

c) Zero Gravity Testing Device

Before testing our Z.R.T.D. with the reaction wheels on board we made sure it functioned on its
own. With this we faced success. With no weight onboard the cap would levitate and rotate with zero
resistance and endlessly until pushed or stopped. The size of the air cushion was observed to be roughly
two millimeters. After demonstrating success with no weight we added dummy loads to determine just
how much weight could be levitated before failure. This weight was discovered to be roughly 10 grams.
All weights up to and including 150 grams rotated and translated smoothly, showing minimal or no signs
of “shearing” or shaking. Above this weight, the air cushion would start to show signs of failure. When
levitating a heavy drill bit, well above 150 grams, the cup would begin to shake violently after only
several seconds of compressed air running through the system. This shaking, at these weights, was
unsustainable, and compromised the ability for the device to be used as intended. Our total payload was
well under 100 grams and presented no shearing issues when testing.

21
3) Full system testing
Testing of the RWS was accomplished using the zero-resistance testing device (ZRTD). The air
cushion created by the ZRTD allows for frictionless rotation on one axis. With no friction acting on the
device, the RWS can be isolated from any forces other than the torques created by the RWS itself.

Figure III-2- Test setup used with zero-resistance device

Figure III-2 shows the system configuration used during full system testing. The system was
assembled in the cup and placed on the ZRTD. Rotation was induced by hand. The system was able to
detect motion and activate the motors accordingly, eliminating the induced rotation.

22
 Figure III-3- RWS during testing on the ZRTD

IV - Summary and Recommendations

1) Summary
The final design of the Reaction Wheel System matches all the technical requirements, as shown in Table
IV - 1. The 5% over budget is due to the fact that the other version of the Electronic Speed Controller,
which was cheaper, was not in stock. In order to respect our schedule and deliver the Reaction Wheel
System on time, we had to buy the more expensive version. Thus, it would have been theoretically possible
to stay under budget.

Table IV - 1 - Comparison to Design Technical Requirements

Requirement Design

Must fit into PC/104 dimensions Dimensions: 52x52x52mm ✔

(90x96mm)

Microcontroller integration Utilized Arduino Uno in testing phase

✔

Sensors to measure and adjust for changes Final design utilizes GY-521
in rotational speed (3 Axis Accelerometer Gyroscope Module) ✔

23
 Must be able to change angular velocity ✔
from 2°/s to 0.1°/s (or less) Motors can produce up to 0.002424
kg*m^2 *rad/sec of momentum in
continuous operation

Motors must be able to reverse - ✔

Overall weight (of all 3) < 150g 74g ✔

Budget < $500 $525 ✘

RWS will be integrated at ± 10 cm from the Compliant with customer specifications ✔

center of mass on the x and y axes.

2) Recommendations

As was mentioned in Section II-2a, the Faulhaber 1509B motor does not have an electronic speed
controller (ESC), so it was necessary to purchase ESCs for each motor. The ESCs were difficult to
implement into the RWS housing due to their size and shape, and they added a significant amount of weight
and cost to the system. Ideally, the motors used in this RWS would have integrated ESCs to reduce the size,
weight, and complexity of the system.
The motors used for this RWS require a 6v power source. Arduino can only output and directly
control up to 5v of power without additional hardware that was not available during testing. As a result, the
motors were not running at full speed during testing, making it difficult to determine if the reduction of
rotational speed of the test platform was due to the RWS, or simply the result of friction. This necessitated
the addition of a pair of 2mm nuts as improvised flywheels to improve the test data.
Adding the flywheels to the RWS significantly increased its performance. It was capable of
stopping rotations more quickly and at lower motor speeds. The motors are theoretically capable of meeting
the performance requirements without flywheels, but it seems that the RWS would be significantly more
capable if flywheels were added. This can be accomplished for minimal cost and requires little modification.

24
V - References
[1] MaI400 RWS. Retrieved November 30, 2017, from https://www.cubesatshop.com/product/mai-400-
reaction-wheel/

[2] Ricardo Filipe Pereira Gomes, “Development of a reliable and low cost miniaturized
Reaction Wheel System for CubeSat applications”, Thesis to obtain a Master’s of Science Degree in
Aerospace Engineering, Page 1-5, April 2016, Print.

[3]“Reaction Wheel Design”, EMFFORCE Operations Manual- Space Systems Product

Development-Spring 2003, Massachusetts Institute of Technology- Dept. of Aeronautics and
Astronautics, Page 2, spring 2003, Print.

[4] S. Nodehin and U. Farooq. “Satellite Attitude Control using three reaction wheels”. An American
Control Conference, Seattle, 2008

[5] Bhanu Gouda, Brian Fast, and Dan Simon, “Satellite Attitude Control System Design Using
Reaction Wheels”, Aerospace Power & Electronics Simulation Workshop 2004, Cleveland State
University, 2004, PowerPoint Slides.

[6] Bill Dunford,“Basics of Space Flight, Section 2: Flight Projects, Chapter 11: Onboard
Systems”, National Aeronautics and Space Administration, Date Accessed 19 November
2017, Web.

[7] “Attitude Dynamics: Disturbance Torques” AERO-423 Chapter 8- Part 1, Texas A&M
University, PowerPoint Slides.

[8] “Applications of Aerospace Techology: Brushless DC Motors”.NASA, Midwest Research Institute,

January, 1975
[9] M.V.Ramesh, J.Amarnath, et al. “Field Oriented Control for Space Vector Modulation based
Brushless DC Motor drive.” International Journal of Advanced Research in Electrical, Electronics and
Instrumentation Engineering IJAREEIE.
[10] https://science.nasa.gov/science-news/science-at-nasa/2001/ast21mar_1, last accessed October 24th,
2017
[11] R.F. Pereira Gomes. “Development of a reliable and low cost miniaturized Reaction Wheel System
for CubeSat applications”.Técnico Lisboa, April, 2016.
[12] R. Joseph Cassady, Roger M. Myers, inventors; Primex Technologies Inc, assignee. Three Axis
Pulsed Plasma Thurster With Angled Cathode And Anode Strip Lines. US patent 6,173,565. January 16,
200
[13] Sidi, Marcel J. “Spacecraft Dynamics and Control: a Practical Engineering Approach.” Cambridge
University Press, 2006.

25
[14] https://www.faulhaber.com/fileadmin/Import/Media/EN_1509_B_DFF.pdf Accessed April 25th,
2018
[15] https://www.faulhaber.com/fileadmin/Import/Media/EN_SC1801S_DFF.pdf Accessed May 8th,
2018

[16] Gavrilovich I., Krut S., Gouttefarde M., Pierrot F., Dusseau L., “Robotic Test Bench for CubeSat
Ground Testing: Concept and Satellite Dynamic Parameter Identification” 2015 IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS) pp. 2- 6.

26
VI - Appendices

Appendix A - Momentum requirement for the x,y and z axis

Z - axis
1 1 5√2 + 5
𝐼 = 2
× 𝑚 × 𝑟2 = 2
×4× 2
𝐼 = 0.007286 𝑘𝑔. 𝑚2
𝜃 = 2 ∘/𝑠 = 0.0349 𝑟𝑎𝑑/𝑠𝑒𝑐
𝐻𝑠𝑝𝑎𝑐𝑒𝑐𝑟𝑎𝑓𝑡 = 𝐼 × 𝜃𝑠 = 2.54 × 10−4 𝑘𝑔. 𝑚2 . 𝑟𝑎𝑑/𝑠𝑒𝑐
1 𝑘𝑔 1 𝑚2
𝐽 = 0.69 𝑔. 𝑐𝑚2 × × = 69 × 10−8 𝑘𝑔. 𝑚2
1000𝑔 10000 𝑐𝑚2

And,
𝐻𝑤ℎ𝑒𝑒𝑙 = 𝐻𝑠𝑝𝑎𝑐𝑒𝑐𝑟𝑎𝑓𝑡 = 2.54 × 10−4 𝑘𝑔. 𝑚2 . 𝑟𝑎𝑑/𝑠𝑒𝑐
𝐻𝑤ℎ𝑒𝑒𝑙 = 𝐽 × 𝜃𝑤ℎ𝑒𝑒𝑙

Thus,
𝐻𝑚𝑎𝑥
𝜃𝑤ℎ𝑒𝑒𝑙 = 𝐽
= 3,685.7 𝑟𝑎𝑑/𝑠𝑒𝑐
𝜃𝑤ℎ𝑒𝑒𝑙 = 586.6 𝑅𝑃𝑀

X and Y - axis

Assumptions:
→ mass of the CubeSat equally distributed
→ Reaction wheel system located at +/- 10 cm from the center of mass
1
𝐼 = 3
× 𝑚 × 𝐿2
1
𝐼1 = × 0.66 × (0.05)2 = 0.0055 𝑘𝑔. 𝑚2
3
1
𝐼2 = × 3.33 × (0.25)2 = 0.069 𝑘𝑔. 𝑚2
3
𝐼𝑇𝑜𝑡𝑎𝑙 = 𝐼1 + 𝐼2 = 0.0055 + 0.069 = 0.0699 𝑘𝑔. 𝑚2
𝐻𝑠𝑝𝑎𝑐𝑒𝑐𝑟𝑎𝑓𝑡 = 𝐼 × 𝜃𝑠 = 0.0024 𝑘𝑔. 𝑚2 . 𝑟𝑎𝑑/𝑠𝑒𝑐
1 𝑘𝑔 1 𝑚2
𝐽 = 0.69 𝑔. 𝑐𝑚2 × × = 69 × 10−8 𝑘𝑔. 𝑚2
1000𝑔 10000 𝑐𝑚2

27
And,
𝐻𝑤ℎ𝑒𝑒𝑙 = 𝐽 × 𝜃𝑤ℎ𝑒𝑒𝑙

Thus,
𝐻𝑤ℎ𝑒𝑒𝑙
𝜃𝑤ℎ𝑒𝑒𝑙 = 𝐽
𝜃𝑤ℎ𝑒𝑒𝑙 = 5,535 𝑅𝑃𝑀

28