Download Technical Paper - Edge - Rochester Institute of Technology

Multidisciplinary Senior Design Conference
Kate Gleason College of Engineering
Rochester Institute of Technology
Rochester, New York 14623
Project Number: P12408
NEAR SPACE SOLAR POWER CONDITIONING MODULE
Stephen Giannotti
Electrical Engineering
Michael Hudak
Computer Engineering
Jacob Emenheiser
Electrical Engineering
Donald Lucas
Mechanical Engineering
William Dorney
Electrical Engineering
Abstract
High altitude ballooning has been a method of scientific research for years and is a growing activity among
electronic hobbyists. These balloons ascend into a near space environment at an altitude of approximately 100,000ft
(30,000m) above the earth’s surface into the atmosphere. Each balloon is typically equipped with communications
equipment, a GPS, and data acquisition hardware. Standard high-altitude balloon flights last less than 24 hours and
are considered short-term flights. The goal of this project was to provide an efficient and reliable power supply
module for a long duration flight high-altitude balloon payload. The Near Space Solar Power Conditioning Module
(NSSPCM) utilizes the availability of sunlight to collect radiant energy and convert it into electrical energy. This
system is designed to be a separate module from a main payload in the high-altitude balloon multi-component
configuration. The module met or exceeded the majority of the design requirements. It supplied the required power,
gathered accurate sensor data, communicated with the mock payload and was contained within a structural sound
enclosure.
Nomenclature
NSSPCM: Near Space Solar Power Conditioning Module
Payload: Module external to NSSPCM. Device that NSSPCM powers and communicates with.
MPPT: Maximum Power Point Tracking: The Genasun solar battery charge controller.
Introduction
Electronics within a high-altitude balloon payload are generally powered solely by batteries. Once the
battery is depleted, there is no other power system available. The NSSPCM project seeks to provide a means of
charging a battery using solar energy to allow for longer missions with lower capacity batteries.
This project was started from members’ work with the successful launch of RITCHIE-1; the RIT Amateur
Radio Club’s High-Altitude Balloon project in May 2011. RITCHIE-1 was designed as a standard short-duration
flight high-altitude balloon payload to provide telemetry data and photographs. The NSSPCM senior design project
was conceived to create a system that could support a second iteration of RITCHIE-1.
Design Process
The NSSPCM design needed to provide a system that can harvest solar energy and convert it to electrical
energy for storage and distribution. Electrical energy from the battery needs to be supplied across a 2m cable to an
external payload. The payload requires a 12V supply with an average load of 2W for three days with a maximum
Copyright © 2012 Rochester Institute of Technology
Page 2
load of 5W for 6 minutes a day. The design also required the ability to monitor system inputs and outputs for data
collection from sensor circuits. A microcontroller allows for system monitoring and control with configured
circuitry and firmware. In addition to the power output, the NSSPCM needed to provide a communication output for
data transfer.
For structural support and protection from environmental conditions, a mechanical structure needed to be
designed. It had to be lightweight, yet robust enough to withstand harsh environmental conditions throughout the
length of the flight and survive a return descent. Module environmental conditions include an external temperature
range of -40ᴼC to 45ᴼC, and the absence of convection due to a vacuum. The module must survive an impact
velocity of at least 7 m/s to ensure adequate structural strength.
Figure 1: NSSPCM High-Level System Design
Solar Energy Harvesting Method
The solar panel chosen had to produce adequate energy during the day to both power theload and charge
the battery. In order to stay within the NSSPCM total weight requirement of 6 lbs, careful consideration was taken in
choosing the type of solar panel. With the potential for adverse weather conditions, it was imperative that the solar
panel be robust. In addition, the minimum solar panel voltage acceptable for this project was 12.5V. The
culmination of criteria lead to the selection of a flexible thin film solar module. Under stress from the environment
these solar panels can bend unlike rigid solar panels which may crack instead. Thin film is lighter weight and does
not require the heavy case of a traditional solar panel. All of these requirements were met by the 1 meter (WSLE0240-24-ST-06-B) semiconductor type copper indium gallium selenide made by Ascent Solar. It has a nominal
power of 24W, voltage (Vmax) of 16.7VDC and weighs 0.75 kg (1.65 lbs).
Maximum Power Point Tracking
Since the system power relies on the output of a solar panel, it was necessary to use a maximum power
point tracking (MPPT) algorithm. Implementing a MPPT method allows for the system to intake the greatest power
possible from the solar panel. For the scope of this project, it was decided to purchase a prefabricated MPPT battery
charger module and integrate it into the system. The product chosen was the GV-5-LI-12.5V Low-Power Solar
Charge Controller made by Genasun. Designing a custom built MPPT algorithm in either hardware or software was
not feasible because of additional complexity that it would add to the project.
Project P12408
Proceedings of the Multi-Disciplinary Senior Design Conference
Page 3
Battery
The combined daily requirements for the NSSPCM circuitry and load was determined to be a charge of
6.78 A-hr at 12V resulting in 81.36 W-hr of energy needed. The NSSPCM was designed to provide power to a load
expecting 10 hours of light and 14 hours of darkness. In order to provide for greater power needs, a large safety
margin of additional energy storage was deemed necessary. Since there were weight restrictions for the NSSPCM
module, the type of battery used had to have a high power to weight ratio. The final choice was a 11.1V nominal
voltage, 10.4 A-hr lithium-ion rechargeable battery pack made by Tenergy. At full capacity the battery will be 12.6V
because lithium-ion cells increase in voltage as they are charged. This specific pack included an integrated
protection board to properly balance each cell.
System Circuitry
The design of the system circuitry was dependant on the power requirements for the internal system and the
payload. To power different components from a common source, 3.3VDC was declared to be the internal voltage
supply. The output voltage supply was already determined from the engineering specifications to be 12VDC. Two
pre-fabricated DC-DC converters were purchased and integrated into our design to provide the regulated voltage
needs. For the 3.3VDC converter, a 2.64W (CC3-1203SF) DC-DC converter by TDK-Lambda was used. To provide
12VDC to the payload, a 20W (RPP20-2412SW) DC-DC converter by Recom was used.
Design of the sensing circuitry to measure voltage and current required the balance of several important
factors. The circuits first had to collect accurate sensor data and deliver it safely to the ADC pins of the
microcontroller. Secondly, sensing of each value required a simple low power solution. Lastly, the sensing circuits
could not interfere with the operation of other circuits in the module.
These requirements were met by using precision sense resistors, high resistance voltage dividers and high
accuracy temperature sensors (LM35). Each of the sense lines leading to the ADC pins of the microcontroller has a
normally open analog switch for protection and a resistor and capacitor for noise filtering. The analog switches are
controlled by the microcontroller, so that the pins will be protected any time the microcontroller does not have
power.
Printed Circuit Board
Since the circuitry used surface mount and through-hole components, a two layer board was determined as
the best option for a printed circuit board. A 6 inch by 8 inch board was designed to provide not only an interface
for the system circuitry, but also a support for the MPPT and battery. The concept was to draw heat away from the
active components in the circuits to keep the battery within the proper operating temperature range. Most of the
bottom side of the board as well as the area under the battery and MPPT act as a combined ground plane and heat
sink. This concept serves as a passive heating method illustrated in figure 2.
As many of the surface mount components as possible were selected to be a size of 1206 to ensure they
were large enough for hand assembly. The PCB layout and fabrication was provided by Elmgrove Technologies
(EGT), a division of Photonamics, Inc. This donation to the project was critical to keep the project under budget.
Figure 2: Printed Circuit Board Conceptual Heat Distribution Diagram
Copyright © 2012 Rochester Institute of Technology
Page 4
All small pitch components were placed first followed by the larger surface mount components and finally
the through-hole components. Initially, the jumpers were not placed on the board so that the circuit could be
powered and tested one section at a time.
Microcontroller and Firmware
The MSP430F2234 was selected because of its low power consumption, 12 ADC channels, and hardware
UART for serial communication. This chip also has the capability of being programmed via a 2-wire spy-bi-wire
interface from the MSP430G2 Launchpad, which is a low cost development board for Texas Instruments' value line
series of microcontrollers. The MSP430F2234 has 512 bytes of RAM and 8 kilobytes of flash memory. The
programming tool, Code Composer Studio, is based on Eclipse open source framework and is free for code sizes up
to 16 KB. The MSP430 firmware is written in C.
The major points of concern when designing the firmware were accuracy of the sensing circuits, the speed
at which the sensors could be updated, and the speed at which the module status could be transmitted to a payload.
The sensing circuits connect to analog switches which prevent damage to the microcontroller in the case that power
is applied to the sensors, but the microcontroller is powered off. When power is applied and the microcontroller
initializes, the switches are closed, enabling the sensors to be read by the ADC. Because the sensor values do not
change at a fast rate, the sensors were read once per second allowing the microcontroller to remain in low power
mode during idle time. All values of the sensors are stored as 16-bit unsigned integers in raw ADC format except for
the temperature values which allow for negative values. Equations are provided for a human readable interface to
convert the raw ADC format into the sensors’ actual values. Fault conditions were defined to indicate mission
critical problems which can be detected and possibly corrected in flight. These include temperature out of range and
low battery fault information sent to a payload for communication with the ground or decision making. During the
design process, the use of the MSP430’s internal temperature sensor and voltage supply sensor was added. The
supply voltage sensor was added to aid in the accurate conversion of raw ADC values to actual values. The firmware
was also designed to be interrupt driven. This design approach allows the microcontroller to take advantage of low
power modes as well as keeping the CPU free to work on needed tasks. The code was also designed to include
multiple code blocks grouped by function. The blocks include clock functions, error checking functions, general
purpose input/output functions, communication functions, sensor functions, and timer functions. Code relevant to
each set of functions was contained within the include file. This allows for more flexibility and clarity of code
implementation.
Communication
A simple and effective communication method was required for the NSSPCM to receive a query from a
payload and transmit a response to a payload. Due to weight concerns of a communication cable, a minimum
number of wires between the NSSPCM and a payload was required. RS-485 at the physical layer was chosen due to
its resistance to noise (differential signaling), the ability to connect multiple transceivers to a communication bus
running between different packages of a high altitude balloon assembly, and it's use only 3 wires. The use of a
MAX3483 chip allowed for low power consumption while still supporting the RS-485 communication standard.
This chip was easily interfaced with the UART of the MSP430F2234.
Structural Design
The main structure of the NSSPCM is constructed of aluminum framing. Aluminum provides the
necessary strength while allowing the module to remain light weight. Aluminum also offers several options for easy
manipulation which made that material an ideal choice. The overall structure of the module is represented in three
basic parts: the enclosure, the cradle, and the panel mount.
The interaction of each part is as follows. The enclosure is an expanded polystyrene case. This case
measures externally 9 inches by 11 inches by 7.5 inches. The enclosure has a uniform thickness of 1.5 inches which
provides the needed insulation to maintain a safe operating internal temperature for the duration of the flight. The
enclosure is encased by the cradle. This network of aluminum strapping (1/16 inch thickness) allows for structural
support of the enclosure as well as convenient means of mounting to other components. These two components
together are attached by four (4) ½ inch hex-head machine bolts to the panel mount. The panel mount is a structure
that outlines the solar panel and provides mounting points to attach the enclosure.
Project P12408
Proceedings of the Multi-Disciplinary Senior Design Conference
Page 5
The Enclosure
The expanded polystyrene enclosure is used to protect and insulate the system electronics. The PCB is
designed to be a snug fit within the enclosure to minimize effects of in-flight turbulence. The enclosure’s thickness
of 1.5 inches provides the necessary insulation to protect the circuitry, and components from both freezing during
the night and overheating during times of direct sunlight contact. To mitigate the effects of heating due to solar
radiation, the enclosure is completely covered with aluminum foil. The foil acts as a reflective agent that will deflect
or reflect the majority of the suns radiation. This effect coupled with the insulation provided and the circuitry design
will allow the internal environment of the enclosure to remain relatively constant. The thickness also provides
significant structural protection that eliminates the need for complete encapsulation using an aluminum structure.
Aluminum strapping was used to provide minimal protection and generate mount points.
The Cradle
The cradle is constructed of 1/16th inch thick, ¾ inch width aluminum strapping. This strapping is easily
manipulated into the needed shapes without extensive or complex machining. The thickness is sufficient as the
frame will be supporting the enclosure in tensile loading while allowing the NSSPCM to remain as lightweight as
possible. The enclosure containing the PCB has an estimated weight of 3.75 pounds and is well within the allowable
limits of the material used. The unit is constructed of three (3) individual pieces of aluminum that were
independently bent to fit the enclosure. The pieces were then welded into the structure pictured. This structure
provides a firm press fit with the enclosure which is then secured in place by the attachment of the panel mount.
Figure 3: Cradle
Panel Mount
The panel mount provides structural support to the flexible cell used. The outer frame is constructed of part
1/8th inch thickness aluminum strapping and part 1/16th inch thickness aluminum. These two parts are used to secure
the panel on the frame. Holes through both the frame components as well as the panel’s edge allow for 20 machine
bolts to be used to tighten the frame onto the panel. These bolts utilize lock-nuts to minimize the risk of in-flight
loss due to vibration. The lower section (1/8 inch thickness) of the panel frame is welded to the shown cross
members. These cross members (marked by blue triangles) are used as mounting point to the cradle. The outer
cross members (marked by red diamonds) provide the mounting locations for the suspension wire that will be used
to tether the NSSPCM to a high-altitude balloon.
Figure 4: Solar Panel Frame
Copyright © 2012 Rochester Institute of Technology
Page 6
Results and Discussion
Figure 5: Assembled NSSPCM
The NSSPCM module structure was successfully assembled and correctly houses all components. The PCB
fitted as anticipated into the enclosure and the lid was held on securely by the frame. During the drop test, the
structure frame suffered significant permanent bending due to the load of the solar panel. All bending was able to be
repaired without use of tools and did not adversely affect the overall integrity of the frame or enclosure. The fully
constructed module weighs a total of 6 pounds, 10 ounces, which falls short of the requirement of 6 pounds.
During assembly, testing and troubleshooting, some changes were required to the PBCA to ensure proper
operation of the circuitry. The silk screen of the component internal temperature sensor U4 was mirrored on
NSSPCM-PCB-001 REV 1.0. The part was placed in the opposite orientation from the silk screen to ensure the
proper electrical connections and operation of the circuit. The output of the 12 volt DC-DC converter is electrically
isolated from the input. The sensing circuitry on the output line was referenced to the output (-). Therefore, a
connection between output (-) and system ground was required so that the sensing circuitry had the same reference
as the ADC of the microcontroller. In the layout of NSSPCM-PCB-001 REV 1.0, the footprint of U6 was incorrect.
The issue was resolved for this build by placing the part upside down on the PCB and connecting it in a dead-bug
style.
A subsystem test was conducted to analyze the amount of energy that could be collected in one day.
Measurements were taken from 7:30 AM to approximately the time when the sun was highest in the sky (solar noon)
12:30 PM. These values were then extrapolated for the latter half of the day. The total energy collection for one day
was calculated to be 118 W-hr which surpasses the required energy collection by over 36 W-hr. This test proves the
solar panel is capable of powering our module.
Figure 6: Solar Panel Test Power Output
The NSSPCM met the power supply requirements of providing 12VDC to an average load of 2W and a
max load of 5W. Engineering specifications require the output be protected from over current. This was to be
accomplished by limiting current draw to 1 amp. During the design, it was decided not to fuse the output and rely on
Project P12408
Proceedings of the Multi-Disciplinary Senior Design Conference
Page 7
the overload protection within the Recom 12V DC-DC converter. The maximum rated output current of the 12V
DC-DC converter is 1.67A and that does not meet our specification of 1A. When testing the overcurrent protection
of the 12V DC-DC converter, a 5 ohm load was connected to cause the output to shut off at approximately 2 amps of
output current. The DC-DC converter however did not shut off and supplied a measured 11.95VDC with a current of
2.29A (27.36W). The test was not continued because the system was already drawing approximately 3A from the
battery which was more than expected to handle from the design. This downfall in design must be accounted for in
future iterations. In its present configuration, all current draw considerations must be accounted for by the payload.
To prove that the module would supply the required power over a 48-hour span, a longevity test was
conducted. The solar panel was placed outdoors and the module was powered-on to supply 12V to the average load
of 2W. The first test run lasted approximately 15.5 hours and was stopped due to adverse weather conditions. A
second test iteration lasted approximately 64 hours. It was observed during both tests that the module adequately
provided power to the load over the course of multiple light and dark cycles. From collected data, it was also
observed that the battery was adequately charged throughout the day to provide enough power during periods of no
sunlight.
The PCB passive heating method was tested by measuring various spots on the PCB every 15 minutes for
two hours while the module was in operation. The test results showed that the heating method worked according to
design in a convection present environment. The temperature measurements showed an increase in temperature from
the start of the test to the end of the test. The most important observation was that the bottom of the battery yielded
the highest temperatures. This result is critical because the battery is the most temperature sensitive device in the
module and requires a strict operating range
The firmware was tested using a custom coded Diagnostic Utility containing a graphical user interface
(GUI) written in Java. The purpose of the Diagnostic Utility was to act as a payload connected to the NSSPCM. The
interface was able to issue commands to the NSSPCM and display results. The internal functions of the program
relating to serial interface were C code from the NSSPCM microcontroller converted to Java code. The equations
used to convert raw ADC values into actual measured values were included in the Diagnostic Utility. The interface
provides a way for a user to interact with the NSSPCM, verify that the communication bus is active, and read the
converted values of the sensors. The interface also reports the time taken from command issue to response returned
for PING and GET_STATUS. All tested response times were below the 1 second specification. The average round
trip time for a PING command is 810 ms and the average time for a GET_STATUS command is 874 ms which is
within specification. The Diagnostic Utility can also save data at specified intervals to a comma separated value file
which is useful for diagnostic data logging. Temperature and voltage sensor values reported by the Diagnostic
Utility were accurate and exceeded specification.
The largest challenge for the communication subsystem was timing. The Diagnostic Utility requires 200 ms
between transmitting a request to the NSSPCM and beginning to listen for a response. This is dependent on the
hardware from which the Diagnostic Utility is run. A manual 200 ms delay was applied in the executeMessage
function to accommodate this. Additional 2 ms delays were required in the transmit Message function to allow for
proper reception of bytes in the Diagnostic Utility. Originally, we had planned on using the MSP430 Launchpad’s
UART to communicate with the NSSPCM. Using this hardware created two problems. The Launchpad’s maximum
rated speed was 9600 baud instead of the intended 19200 baud. Because the communication speed still remained
within specification, the timeouts and UART settings were reconfigured for 9600 baud. A second issue with the
Launchpad was the lack of a request to send (RTS) signal from the UART which would enable the driver of the
MAX3483 circuit and allow the Diagnostic Utility to transmit a request then return to receive mode. This was solved
by propagating the transmit signal from the Diagnostic Utility through an MSP430G2231 microcontroller to the
MAX3483 chip. The MSP430G2231 would detect the start bit of the UART transmission and enable the RS-485
driver for the duration of the transmission. The ability of the NSSPCM to transmit alerts to the payload was disabled
due to communication conflicts. The Diagnostic Utility would cease to function if it attempted to transmit a query to
the NSSPCM while it was simultaneously receiving an alert from the NSSPCM. This occurred because a collision
avoidance mechanism was not implemented in the Diagnostic Utility communication code. Because alerts were not
a specification, they could be disabled safely and the Diagnostic Utility continued to have the ability to receive fault
information by polling the NSSPCM.
Copyright © 2012 Rochester Institute of Technology
Page 8
Conclusion and Recommendations
There are several issues that would need to be addressed in a second iteration of this project. First, the
following errors with the PCB need to be addressed: correct the silk screen of the internal temperature sensor U4,
connect output(-) and system ground with a trace and correct the footprint of the 3.3V DC-DC converter U6. In
order to allow another step in the circuit testing it would useful to add a jumper at the output of the 3.3V DC-DC so
that the microcontroller circuits can remain unpowered while the power path is being tested. Finally, the connectors
on the PCBA are very convenient for testing and troubleshooting, but they should be removed for dependability
during flight.
Regarding the mechanical structure, the next iteration it would be recommended that support braces are
added to stiffen the junction between the cradle portion of the frame and the solar panel frame. Thicker aluminum
stock would also eliminate a significant amount of deflection, but would dramatically increase the weight of the
module. Using a more shallow enclosure will decrease the amount of building materials and in turn decrease weight.
For future firmware development and testing, it is recommended that a USB to RS-485 converter be used
instead of the MSP430 Launchpad. Using the converter, a higher baud rate can be achieved as well as eliminating
the need for an additional microcontroller, driver chip and breadboard. Also, it is recommended using a different
serial interface other than the javax.comm library. This library is no longer supported by Oracle and requires files to
be installed on the computer to interface with the serial port. It would also be useful to add collision avoidance to the
communication code of the Diagnostic Utility.
To better test and investigate the NSSPCM passive heating method, a vacuum chamber needs to be used to
eliminate the effect of convection. A vacuum chamber is needed to properly simulate the harsh conditions of a near
space environment.
In order to avoid damaging the payload, special care must be taken to limit the output power to within an
acceptable range. This may be accomplished by replacing the 12V DC to DC converter or implementing a fuse. It is
possible that the heat sink of the NSSPCM circuit board interfered with the operation of a thermal cut-off within the
DC-DC converter and caused the device to stay on by dissipating the converter heat throughout the board. Further
tests need to be conducted on the device independently to determine the cause of these results.
After design, building, and testing the NSSPCM, it was determined that the device functioned properly and
met the majority of the requirements. A second iteration of the project can take note of the specifications that were
not met and further develop the design.
Acknowledgements
Faculty Guides
Vince Burolla
Leo Farnand
Dr. Dorin Patru
Additional Support
Gordon Davies
Project Sponsers
RIT Multidisciplinary Senior Design Department
RIT Electrical Engineering Department
Elmgrove Technologies
Special Thanks
Jeff Lonneville – CEMA Laboratory Manager
Elle Canales – Logo Designer
Lowes Henrietta
Bergquist Company
Project P12408