39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit AIAA 2003-5246

20-23 July 2003, Huntsville, Alabama

AIAA-03-5246

LOW COST THRUST VECTOR CONTROL SYSTEM DEVELOPMENT

FOR SMALL LAUNCH VEHICLES

Stanley Baksi*, Eric Besnard† and John Engberg

Mechanical and Aerospace Engineering Department
California State University, Long Beach

John Garvey‡ Layne Cook and Craig Trent

Garvey Spacecraft Corporation Universal Space Lines, LLC
Huntington Beach, CA Newport Beach, CA

successfully static fire tested on Sept 22nd, 2002.

Abstract
P-3 was flown in open loop on Feb 22nd 2003 and
While many organizations have been involved demonstrated in-flight thrust vectoring. At least two
in the development of nanosatellites, relatively few additional flights are planned with P-3 to complete
of them have focused on the development of small the development and test of the guidance,
launch vehicles (SLVs) dedicated for such navigation and control as well as telemetry
payloads. One of the key capabilities for such systems.
SLVs is cost-effective thrust vector control. This
1. Introduction
paper presents the status of an on-going effort
addressing this requirement which is being For the numerous academic institutions and
implemented within the framework of the California other organizations developing small satellites
Launch Vehicle Education Initiative (CALVEIN). around the world, obtaining a launch has often
The CALVEIN team, which is based on a proven to be the primary obstacle to success. On
cooperative partnership between California State the nanosatellite scale (unofficially defined as 1 to
University, Long Beach (CSULB) and Garvey 10 kg) only a handful have been launched [1] and
Spacecraft Corporation, plans to demonstrate none in the picosat (0.1 to 1 kg) and femtosat
initial TVC concepts by using the school’s (<100 g) ranges. This is primarily due to the fact
Prospector-3 (P-3) prototype vehicle to perform a that launches are severely restricted by high costs
pitch maneuver and return to straight flight. Electro and lack of dedicated launch vehicles. Launching
mechanical actuators (EMA) gimbal the 500 lbf small satellites as secondary payloads usually
engine in two axes for pitch and yaw control. An entails long lead times between completion of the
onboard computer processes inputs from the three hardware and actual flight due to integration and
rate gyros and commands the EMA’s to guide the regulatory issues. Developing a small launch
rocket along the predetermined flight path. vehicle dedicated to launching nanosatellites will
Analysis of the vehicle dynamics and control enhance responsiveness and allow many new
algorithm shows that the TVC system developed is small payloads to be developed and deployed.
stable with adequate margins. The system was

*
Graduate Student, AIAA Member
†
Associate Professor, AIAA Member
‡
AIAA Senior Member

Copyright © 2003 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
 California State University, Long Beach 2. Prospector rockets as prototype SLV’s
(CSULB) and Garvey Spacecraft Corporation Students from CSULB in collaboration with
(GSC) have been developing small prototype GSC and its affiliates, under the framework of the
launch vehicles with the objective of eventually California Launch Vehicle Education Initiative
evolving their system to a small launch vehicle (CALVEIN)*, have been developing the Prospector
(SLV) capable of delivering 10 kg to Low Earth rockets. The Prospectors, follow-on to GSC’s
Orbit (LEO). These prototype vehicles are series of Kimbo vehicles, are liquid propelled
described in Sect. 2. Development of a thrust rockets, using ethanol and liquid oxygen, have a
vector control (TVC) system specifically designed length between 3.65 m (12 ft) and 4.25 m (14 ft), a
for SLV is critical for this purpose. After reviewing diameter of about 25 cm (10 in) and a lift-off mass
several candidate approaches, the team then between 68 kg (150 lb) and 91 kg (200 lb). This
selected engine gimballing, which allows for pitch size puts it at about half the size of an actual
and yaw control. Roll, which could be nanosat SLV. The Prospector rockets serve as
accomplished by the addition of another engine or test beds for prototype development of
with micro-thrusters, will be addressed in the technologies like avionics, propulsion system, and
future. TVC, which can be evolved to small launch
A simplistic representation of a TVC system is vehicles capable of delivering payloads in orbit.
shown in Fig. 1. Along each axis, the measured Prospector-1 (P-1) and Prospector-2 (P-2) were
angular rate of the vehicle is differenced with the both flown and recovered in 2001 and 2002,
desired rate and an engine deflection command is respectively (Fig.2). A payload bay was added to
produced, which applies a moment to correct the Prospector-2 to fit a small test payload developed
error. by students from the Space System Development
actual Lab at Stanford University. This latter vehicle is to
Engine endgine
deflection Actuator deflection Airframe be reused for the flight of an aerospike engine
comand dynamics
developed by CSULB and successfully static fire
pitch rate tested in June 2003 [2] and more payloads will be
Rate
Engine deflection
sensor flown. More on the program can be found in Refs.
command
compensation
2 and 3.

Fig. 1. Simplistic representation of Thrust Vector Control

along one axis, e.g. pitch

This paper presents the various components

and subcomponents of the blocks shown in Fig. 1.
TVC is demonstrated in the prototype small launch
vehicles developed jointly by CSULB and GSC,
the Prospector/Kimbo rockets. These are briefly
presented in the next section. Sect. 3 presents the
requirements set forth for the development of the
TVC system. It is followed by Sections 4 and 5 in
which the gimballing mechanism and avionics
system are presented, respectively. Once these Photo by Tony Richards
(a) (b)
are known, the gimballing mechanism (Sect. 5)
and avionics can be shown (Sect. 6). With these Fig. 2. Prospector prototype launch vehicles in flight, (a) P-
elements in place, Sect. 7 investigates the control 1 and (b) P-2
system characteristics and is followed by an
overview of test conducted so far and tests
planned for the next few months in Sect. 8.
*
CALVEIN is hosted by CSULB and has the primary goals of providing
hands-on hardware experience to future aerospace engineers, while
also conducting launch vehicle research and development

2
 Prospector–3 is the third rocket in the series. It based on data available from the basic Prospector
has the characteristics of P-1 and P-2 and is being configuration.
used to develop the TVC described in this paper. A simple solid model of the vehicle was made
The first flight test of Prospector-3 took place in and used to determine mass of the vehicle, center
Feb. 2003 and included payloads from the of mass and moments of inertia. Also, vehicle
University of Southern California’s (USC) Microsat aerodynamic properties, i.e. center of pressure
project as well as from JPL’s MEMS’ laboratory. and pitching moment derivative, Cmα, were
determined using a simple tool [4].
The results of the analysis were
3. Requirements and airframe dynamics
1. Mass of fueled vehicle = 87 kg
2. Position of center of mass at 2458.46 mm
3.1. Requirements from the nose and along the centerline of
TVC allows a vehicle to perform maneuvers the rocket
and deliver payloads in LEO and is also essential 3. Moment of Inertia about pitch and yaw
axis = 127 kg m2
for increased fuel mass ratio at lift-off. To
4. Center of pressure 584.2 mm below the
demonstrate TVC development on the prototype center of mass
SLV, it was decided to require P-3 to fly along a 5. Pitching moment derivative, Cmα = -0.44
simple and predetermined path. The trajectory of
the vehicle, illustrated in Fig. 3, is defined as,
4. Gimballing Mechanism
• At T+ 3 s the rocket will perform a
negative pitch maneuver that increases its TVC is achieved here by gimballing the engine
attitude to 30o, where T is takeoff time. about the pitch and yaw axes. As shown in Fig.4,
• At T+ 5 s, command a turn that decreases a ring is designed to act as a universal joint.
the attitude to less than 5o.
This trajectory is shown for a test being conducted
at the Mojave Test Area (MTA), a site owned and
operated by the Reaction research Society where
all CALVEIN tests have been performed to date.

P-3 trajectory NORTH

o
θ3 < 5 (a)

o
θ2 > 30
Ground track
of trajectory o
θ1 < 3
T+4
Launch rail
T+2
Blockhouse

Observation
Bunker (b)

Fig. 4. Gimballing mechanism: (a) solid model and (b)

prototype
Fig. 3. Flight path for demonstrating TVC
The system has linkages attached to electro-
3.2. Airframe dynamics mechanical actuators (EMA) to rotate the engine
In order to derive requirements for the TVC about the pitch and yaw axis. The decision was
system, the team estimated airframe dynamics made to start with this approach instead of direct
linear drives because of perceived advantages in

3
prototyping. The EMAs are DC servomotors that 3 are commercially available servomotors and are
are coupled to the engine by gear train and shown in Fig. 4b. It is anticipated that future
linkages. EMAs are specially well suited for use in iterations will incorporate linear drive motors.
small launch vehicles because of their low power
requirements. The use of EMA also eliminates the
risks of leaks present in hydraulic and pneumatic
systems and keeps the system simple and easy to
control using commercial-off-the-shelf (COTS)
electronic components.
The power requirement of the motor was
estimated by simulating the pitch attitude of the
vehicle in open loop control. For this analysis, the
vehicle mass and velocity were assumed
constant. Only one-axis rotational dynamics was
considered. Fig. 5 shows the nomenclature used
for describing attitude of the vehicle (θ) and engine
(a)
deflection (δ).

(b)
Fig. 5. Nomenclature for analysis of P- 3
Fig. 6. Vehicle response (continuous line) to sine engine
Fig. 6 shows the vehicle response in pitch due o
deflection (dotted line), (a) ±3 engine deflection with low
to engine deflection, with low and higher airframe o
airframe damping and (b) ±5 engine deflection with larger
damping. In general, a low engine deflection of 3 airframe damping
to 5 deg. (dotted line) is sufficient to cause a
vehicle deflection (continuous line) close to 30 8

deg. 6

This analysis shows that it is sufficient to

4
Power2
design the gimballing system for ± 5o engine (W) 0
deflections. The analysis also shows that, with -2
0 0.2 0.4 0. 6 0.8 1 1. 2 1. 4 1. 6 1. 8 2

little damping, as is expected here, the vehicle -4

exhibits a fish tailing attitude which must be -6

-8

eliminated by the control system. It demonstrates, - 10

therefore, that a closed loop control system must T im e

be used to control the vehicle.
The power requirement corresponding to the Fig. 7. Power requirements for actuators
maneuver of Fig. 6b is shown in Fig. 7. Although
suitability of EMA will ultimately depend on the
bandwidth of servomotors, this calculation gives a
minimum requirement. The motors selected for P-

4
 5. Avionics Nose Cone
The avionics system used to integrate the
components of Fig. 1 is based on existing
commercially-off-the-shelf products. The flight
Longeron
computer operating system is Linux and a C-
program sends commands to valves and servos
and receives data via the RS-232 port.
Ethanol Bulkhead
The TVC data received by the flight computer tank
includes engine position in two axes (from servo-
control boards connecting the servos), angular
rates (from rate gyros positioned along all 3 axes), LOX tank
and accelerations to determine vehicle flight path.
Plans also include the addition of a GPS sensor to
complement accelerometer data. Although not
controlled in roll, the vehicle roll angular rates are Fins
sensed to be able to control the trajectory of the
vehicle. Fig. 8. Vehicle structural components
Telemetry data is down-linked at a lower rate Fig. 9 shows the first bending mode of the
than what is used in the control loop using rocket, which occurs at 12.7 Hz, using Lancoz
standard wireless internet technology with a 1-W algorithm.
signal amplifier on both rocket and ground.
6.2. Analysis of the TVC linkage alone
6. Vehicle structural dynamics
The next step in the structural dynamics
For flying the vehicle along a predetermined analysis involves the modeling of the TVC linkage
path, it is important to understand the impact of alone. Fig. 10 shows the components of this TVC
the vehicle elastic properties on the stability of the linkage, which then will be attached to the vehicle
rocket. This information is used for proper base plate.
placement of the rate gyros to minimize the impact The following assumptions were made for the
of the bending dynamics and also for the design of analysis:
the filters. • Shell elements: Rocket Base Plate,
The natural bending modes of the rocket were Universal ring, Engine Top Plate, Block,
determined using the FEM package in IDEAS [5]. Bearing, Crank Ring, Gear box block
The first analysis was performed for the vehicle • Beam Elements: Actuator interface, Rod
structure alone, then for the gimballing system Ends, Actuator shaft, Actuator motor, Rod
alone, and then for the combined system. Ends (Connecting the Crank Ring to the
6.1. Structural analysis of the vehicle without the Actuator Interface), Actuator motor
engine • Spring Elements: Gear stiffness
• Boundary Conditions:
Fig. 8 shows the structural components of the 1. The bolt holes for the longerons on
vehicle. The assumptions made for determining the rocket base plate were fixed
the natural modes of the rocket were: 2. The nodes representing the rod ends
• Shell elements : Rocket base plate,
had the rotational degree of freedom
Bulkheads, Brackets connecting bulkhead
to longerons, Rocket shell released
• Beam elements: Longerons 3. Node representing the sensor
• Rigid Elements: Connection of longerons (encoder) is coupled to the node
to the bulkheads representing the actuator shaft by
• Boundary Conditions: free – free rotation along the z axis and the
coefficient = -0.5

5
 Fig. 9. First bending mode of the rocket (12.7 Hz) Fig. 11. First bending mode of TVC linkage

7. Control system for TVC

This section describes the integration of the
elements of the control system, analyzes its
stability and response, and presents simulations
for the desired flight profile described in Sect. 3.
7.1. Control loop
The elements of the closed loop control syst of
Fig. 1 are shown in detail in Fig. 13 as derived
from Ref. 6. The flight computer commands the
vehicle attitude (θCMD). This command is
Fig. 10. TVC components differenced with the measured attitude (θ), and the
Fig. 11 shows the first bending mode of the error is gained to produce a rate (angular velocity
TVC linkage, which occurs at 139 Hz. of turn) command. The rate command is
differenced with the measured rate (using a solid
6.3. Combined Model
state rate gyro), and the error is gained to produce
The final step in the entire vehicle structural an engine deflection command (δ). The gain
dynamics characterization involved combining estimates for rate and attitude are based on
both models. The TVC linkage was attached to the previous data and analysis.
rocket body. The combined model gave the first
rocket-bending mode at 12 Hz and the first TVC
bending mode at 101 Hz, as shown in Fig. 12.

6
(a) (b)

Fig. 12. First bending mode of vehicle/TVC link combination (a) for vehicle and (b) for the TVC linkage with engine

Commanded Vehicle
time Deflection Deflection

Actuator

Vehicle

Fig. 13. Detailed control loops for TVC of P-3

7
 The industry standards for gain and phase
7.2. System stability and response
margins are 6 dB and 30 deg., respectively [7].
This section addresses the stability of the Usually a good gain margin corresponds to a good
system developed with the assumptions described phase margin. In case of P-3, however, the phase
above. In addition, several other effects needs to margin is slightly below the 30o although the gain
be investigated, namely engine inertia (giving rise margin is adequate. Prior to flying the closed loop
to “tail-wags-dog” – TWG), propellant slosh and system, compensation techniques will be
actuator backlash. investigated to increase the phase margin.
Forces due to propellant sloshing in the tanks 0.2
can destabilize the vehicle. The slosh mode
frequency is a function of the acceleration acting 0.1

on the vehicle and the radius of the propellant

0
tanks. For the current vehicle, the slosh frequency
will vary between 2.2 and 2.4 Hz. The smooth wall

Imaginary Axis
-0.1

damping ratio of the tanks is approximately 10-4, so

-0.2
the time constant (time for the response to reach
63% of its steady-state value) is 663 seconds. -0.3

Since the engine burn time is only 10-15 seconds,

slosh disturbances are not a concern. -0.4

The engine inertia effect manifests itself in the -0.5

presence of a pair of zeros on the imaginary axis -4 -3.5 -3 -2.5 -2 -1.5
Real Axis
-1 -0.5 0 0.5 1

of the Nyquist plot, which limits the response of the

autopilot. The frequency at which these pair of Fig. 14. Nyquist plot for control loop
zeros occurs at is called the “tail-wags-dog” (TWD)
frequency. For the current vehicle, it is around 9
Hz, which is well above the autopilot bandwidth, so
that its effect, although included in the analysis,
was negligible.
The system stability and response was
analyzed with and without actuator backlash. Fig.
14 shows the Nyquist plot for the control system of
Fig. 13 (3 control loops: actuator, vehicle bending
and outer control system) without backlash. Table
1 presents the corresponding data and Fig. 15
shows the control system response to a step input
for engine deflection without backlash.
Margin Frequency
Fig. 15. Response to step input without actuator gear
Low Frequency
backlash
Gain Margin 12 dB 0.9 Hz
(airframe property) The effect of actuator gear backlash was
High Frequency investigated by repeating the simulation with an
Gain Margin 39 dB 10 Hz actuator gear backlash of ±0.5 deg. While this
(bending property) backlash did have some effect on the phase margin
Phase Margin 25 deg. 1.8 Hz of the system, this effect could be compensated by
Bending Mode increasing the servo bandwidth. In addition, the
Margin 5 dB 12 Hz actual gear backlash for the actuator selected here
(phase stabalized) and shown in Sect. 4 was measured at ±0.1 deg.,
Table 1. Gain and phase stability margins (with bending which suggests that the effect of actuator backlash
and TWD effects) without backlash should be minimal and should not lead to system
instability.

8
7.3. Control system simulation described in Sect. 8.1. The next milestone involves
demonstrating in-flight thrust vectoring. This test is
In order to assess the predicted capabilities of
discussed in Sect. 8.2 and future flight tests are
the TVC system to meet the requirements of the
discussed in Sect. 8.3.
flight profile, simulations were performed using the
Simulink module in MatLab [8]. Fig. 16 shows the 8.1. Static Fire Test
commanded attitude. Fig. 17 shows the simulated
The gimballing system was successfully static
actual vehicle attitude along with, in Fig. 18, the
test fired on Sept. 22nd, 2002 at the Mojave Test
necessary engine deflection history. It shows that
Area (MTA), site owned and operated by the
the vehicle has reasonable response time and a
Reaction Research Society (RRS). The system
suitable settling error.
was tested for deflections about both pitch and yaw
Based on this simulation, it is expected that the
axes with a 500 lbf engine using LOX and ethanol
TVC system will be able to meet the requirements
as propellants.
set forth in Sect. 3.
Fig. 19 shows stills from the static fire test
including when the engine reached maximum
θCMD commanded deflection, i.e. δ = 5o. The test also
(deg.) demonstrated the use of the COTS electronic
components for controlling the actuators and
validated the initial software development.

Time (s)

Fig. 16. Commanded vehicle deflection, θCMD

θ (deg.)
(a) (b) (c)

Fig. 19. Static fire test with (a) 5 deg. engine deflection in
Time (s)
pitch, (b) no deflection and (c) both pitch and yaw engine
deflections at 5 deg.
Fig. 17. Actual predicted vehicle deflection, θ
8.2. In-flight thrust vectoring demonstration
The second milestone involved flying the
δ (deg.)
Prospector-3 vehicle (Fig. 20) in open loop with
Time (s)
thrust vectoring about the pitch axis. An engine
deflection of ± 2 deg. was preprogrammed. Three
seconds after lift-off, the engine was deflected to -
2o for half a second, then swung to +2o for another
Fig. 18. Engine deflection required, δ
half second, and returned to 0o. Each maneuver
8. TVC demonstration (i.e. time during which the engine rotates) was
designed to last about half a second.
A series of static fire and flight tests has been The flight test took place on Feb 22nd 2003 at
planned leading to a flight test demonstrating the the MTA. No rate gyro data was collected on this
full capabilities of the TVC system presented here flight. Three seconds after lift-off (Fig. 21), the
to meet the requirements of Sect. 3. The first vehicle successfully thrust vectored. Due to the fact
milestone was to qualify the mechanical aspects of that the airframe has three fins, however, the axial
the system under static fire test. This test is asymmetry induced a roll which prevented the
9
vehicle from coming back to, or close to, the computer, rate gyros, accelerometer, etc.) and with
vertical on the second maneuver. At this point the data being collected and downlinked. It is planned
vehicle was almost horizontal and the recovery to close the control loop in a subsequent flight.
system automatically deployed for a successful
9. Conclusion
recovery.
In addition to demonstrating the in-flight thrust The paper reports on the preliminary
vectoring capability, this test also included the development of a low cost thrust vector control
flight of a payload developed by University of system which could be evolved for Small Launch
Southern California students as part of the Vehicles capable of delivering nanosatellite-type
Microsatellite Project. In addition, several samples payloads to Low Earth Orbit. The system is being
testing a new bonding technique developed by the demonstrated onboard a small prototype liquid-
Jet Propulsion Laboratory for propelled rocket by a series of flight tests leading to
microelectromechanical systems (MEMS) a flight of the closed loop control system.
spacecraft microthrusters were flown onboard P-3 All the elements of a guidance, navigation and
for verification in a flight environment. control system are integrated, as well as telemetry.
Rate gyros and accelerometers, complemented
with a GPS sensor, constitute the elements of the
inertial measurement unit. Pitch and yaw control is
achieved by gimballing the main engine. The paper
shows that the system is stable and responds
suitably to the desired commands.
Tests conducted thus far include a static fire
test and an in-flight thrust vectoring demonstration.
At least two more flight tests are planned, the first
with all components of the TVC system, but without
closing the control loop, and the last one with the
Fig. 20. Engine gimballing system while P-3 is being
loop closed. This tests are planned for late 2003,
readied on the launch rail
early 2004.
References
1. Surrey Satellite Technology Limited, Small
Satellites Home Page,
[Link] 2003
2. “CALVEIN Web Site”,
[Link]
2003
3. Eric Besnard and John Garvey, ‘Educating
Tomorrow’s Aerospace Engineers by
Developing and Launching Liquid-Propelled
Rockets,’ IAC-02-P.1.05, 53rd International
Astronautical Congress, World Space
Congress, Houston, TX, 2002
4. H. A. Toft, “Aerolab”, available at
[Link] 2003
Photo by Tony Richards
5. “I-deas”, product information available at
Fig. 21. Lift-off of P-3 during the first P-3 flight test [Link] 2003
demonstrating in-flight thrust vectoring 6. Greensite. Arthur L, “Analysis and Design of
Space Vehicle Flight Control System”, Spartan
8.3. Future flights Books, New York, 1970
7. Nagrath. L and Gopal M., “Control systems
The next milestone in the TVC flight
engineering”, New York : Wiley, 1982
demonstration program involves performing
8. “MatLab and Simulink”, product information at
another flight in open loop control, but this time
[Link] 2003
with all elements of the system in place (flight

10