PROJECT A Raspberry Pi Weather Balloon

A Raspberry Pi
Weather Balloon
Abstract
The aim of this project is to collect environmental data and take photographs from the upper atmosphere as an
exercise in engineering and programming. The intention is to fly a helium-filled weather balloon up to a height of
37,000 metres, suspending a capsule of sensors powered by a Raspberry Pi microprocessor. Flight path will be
determined using GPS data taken at regular intervals, with a pressure meter to help determine altitude. Tem-
perature will also be taken at the same intervals via another module. Finally, a camera module will be attached
to take a time lapse of the ascent. All of the data will be saved to an SD card. A radio module will broadcast the
GPS signal so it is possible to find it once it has completed its ascent and fallen back to the ground. The cap-
sule will consist of a high density closed cell foam box to protect the Raspberry Pi and sensors from damage
during the fall.

Funding Statement Coding and Raspberry Pi

Royal Society Partnership Grant Since the Pi must conduct multiple tasks, the
components of the capsule are split into modules,
each of which could be developed on multiple Pis.
Introduction

T
This solves potential issues like wiring conflicts and
he aim of this project is to release a
makes bug fixing easier minor coding or wiring
microprocessor-powered data-recording
modifications can cause lots of problems, but any
capsule into the upper atmosphere, attached
work done on one set of modules will not affect the
to a helium-filled weather balloon. This launch has
others. The grouping of the modules is based on
not yet taken place, but much work has been done
which components are dependent on each other.
on the capsule. As the team had experience with
For example, the temperature and pressure sensors
using a small single-board Raspberry Pi computer
are configured using the same Pi, as data will be
(also referred to as Pi), this was used as the
logged to the same text file. The temperature and
microprocessor. Temperature, air pressure and the
pressure file will be completely independent from
GPS data will be recorded at regular intervals during
image files, so the camera can be part of a separate
the ascent using sensors wired to the Raspberry
module. Setting up the workflow like this means that
Pi, and the data broadcast to the ground via a radio
multiple people can code at once on separate Pis. All
module. This will also allow the calculation of the
the scripts that are created this way will be collected
balloons altitude and its location as it returns to
together and run by a single master script, making
Earth. A small camera is also fixed to the capsule to
debugging a lot easier. Each development Pi and
record the flight. Preliminary research into similar
its corresponding breadboard of components will
projects indicated that the balloon would reach
be combined to a single 256MB Pi, to which a high
an approximate height of 37 km, at which point
data capacity SD card with all the scripts and master
the balloon should burst, releasing the capsule.
script on it will be inserted.
The Raspberry Pis descent will be controlled by a
parachute and the impact of landing lessened by a
All of the functions and data collection from various
shock-resistant casing of some kind.
sensors need to be programmed so the Pi can
carry out the desired functions exactly. Python, a
Method programming language that comes preinstalled on
A standard helium-filled weather balloon will be Raspbian (Raspberry Pis operating system), was
used and the capsule containing the Pi and modules chosen for the project, as it is relatively easy for less
attached to it. The dimensions of the balloon will be experienced programmers to pick up. As most Pi
determined based on the mass of the final load. users also use Python, there are many tutorials and
existing codes on the internet that are useful for this
project.

[vitaminoc.com]
20 WWW.YSJOURNAL.COM I ISSUE 17 I ROYAL SOCIETY SPECIAL
 A Raspberry Pi Weather Balloon PROJECT
The first element the project tackles is recording
the ambient environment of the capsule and storing
this data in a text file, beginning with the connection
of the physical sensors to the Pi so they could
communicate. A Raspberry Pi has two rows of 13
General Purpose Input/Output (GPIO) pins that can
be used to connect the Pi to anything that does not
have a USB. Sensors must be attached to these.
Research into how the components should be
attached is summarised in figure 1.

The BMP085 pressure sensor (Figure 2) is much

more complex, consisting of its own small circuit
board. However, it does have labelled pins which can
be easily be connected to the right parts of the Pi.

After accessing the right GPIO ports for each

sensor and calling the right part of the output data, Figure 2: The BMP085 pressure sensor.
temperature and pressure, it is output to a single text After connecting the GPS to the correct GPIO pins,
file. The programme can also be looped and a time the Adafruit website became very useful in helping
stamp added to each reading to aid interpretation. to extract the latitude and longitude from the data
the module provided. The data could then easily be
The next challenge is integrating the GPS module inserted into the temperature and pressure text file.
(Figure 3) into the other sensors and writing the A minor setback faced whilst trying to work with the
capsules latitude and longitude to the same text file. GPS unit is that in order to test and troubleshoot
The GPS module is another stand-alone component the code, a GPS signal was required which was not
with comprehensive pre-installed software that
outputs a range of data from location to velocity.

Figure 3: Adafruit ultimate GPS module

with aerial. www.learn.adafruit.com

always available within the school building.

There is also the possibility of using GPS to calculate

altitude, though further testing and research would
be needed to confirm if it would work above a certain
altitude.

Figure 1: Diagram showing the pins on

a on the Raspberry Pi and connections Casing
to the temperature sensor. The DS18B20 For the casing of the Raspberry Pi and the modules,
temperature sensor is a simple compo- high-density closed-cell foam will be used. There
nent, connected to the 3.3V, GND and were concerns about the impact of low pressure
4th GPIO pin. on the foam and whether this would cause it to
ROYAL SOCIETY SPECIAL I ISSUE 17 I WWW.YSJOURNAL.COM 21
 PROJECT A Raspberry Pi Weather Balloon

Figure 5: Radiometrix NTX2 Transmitter

Figure 4: Capsule (unfinished) with the

battery pack and Pi in place.
expand, possibly resulting in structural damage
and an inability to hold all the components safely.
However, tests on the foam in a vacuum jar showed
that the foam was not affected by the extremely
low pressure in any way and retained its strength on
repressurisation.

The Raspberry Pi and other components were

measured to determine appropriate dimensions that
would leave enough space for all the components.
The foam was cut using a scroll saw and will be
finished with the belt sander to ensure straight Figure 6: YAESU Transceiver
The difficulty was getting the Pi to use this
edges. The casing can be seen in figure 4.
transmitter to communicate with a laptop at a
The casing will be finished at a later date after
distance. The language Radioteletype (RTTY) was
bringing all the different components together onto
used for transmission, as it can convert data into
one Raspberry Pi. It will also be drop-tested with the
basic binary that can be received and easily decoded
small parachute. After this, any modifications for
into the relevant data. Thanks to the University of
strength or fit, such as adding a space on the outside
Southampton, who procured a transceiver that
for the camera, will be added. This will be done later
could receive the signal and output data to the
as all of the modules need to be brought together to
computer (Figure 6), the laptop is able to receive
determine how it will all fit, before foam inserts are
at the frequency the Pi is broadcasting. Weeks of
added to ensure snug, gapless fit, so nothing can be
deliberation and research into which of the many
damaged.
output ports on the transceiver should be used,
involving laptop chipsets and complex specifications,
Data Transmission showed that an audio jack would work perfectly.
The broadcast of data back down to Earth has
been one of the biggest challenges. The NTX2 At this stage, the 434.075MHz radio frequency
transmitter from Radiometrix (Figure 5) is small, light from the transmitter could be converted into an
and, crucially, broadcasts on a frequency outside audio format, and outputted to a computer via the
that of any radio station, removing the likelihood microphone port. To interpret the audio, Dl-FLdigi, a
of encroaching on any regulations. Connecting downloadable interpretation programme that even
the transmitter to the Pi was a straightforward came with a high-altitude ballooning edition, allowed
procedure. us to take this signal and translate it into readable
numbers. The programme required a programming
A laptop was used to receive and interpret the data language, baud rate, frequency band, and the desired
rather than another Pi, as this would require much frequencies to be set. It would then convert any
more processing power than available on a Pi, and sound the laptop picked up to a string of letters and
there are no weight limits. This also allows the numbers.
broadcast of the minimum amount of data from the
Pi, as the data can be processed into graphs on the The next and most challenging task so far is to
ground.
22 WWW.YSJOURNAL.COM I ISSUE 17 I ROYAL SOCIETY SPECIAL
 A Raspberry Pi Weather Balloon PROJECT
code the Pi to convert a text file to RTTY, then In Python the for loop repeats a user-defined
broadcast this encoded files 1s and 0s as high and number of times. In the loop, the picamera.capture
low frequencies. To demonstrate control of the function was used, which was imported at the start
transmitter, with assistance from a technician, the of the script. The function being used takes a photo
GPIO pin, to which the transmitter was connected, and then names it x.jpg in any chosen location, x
was turned on and off. This involved a simple code being the iteration number of the loop. This means
consisting of a loop which set the pin to on (true), that no pictures will be written over, and we are able
waited half a second, then set it to off (false), then to set location allowing multiple time lapses to be
waited before repeating. This would produce a recorded during testing and be stored in separate
regular oscillation between two frequencies, which folders. After the picamera.capture has completed
could be received as two distinct pitches through the loop moves onto the time.sleep function which
the transceiver. However, work on how to encode pauses the code for a set amount of time which can
and transmit the initial text file is still ongoing. It is a be controlled by changing the variable. Once time.
matter of finding the right code on the internet and sleep has completed the code the loop ends and
adapting it to the needs of the project. starts again from the beginning until the code is told
to stop or until the loop has reached the end.
Camera
There will be a camera on the capsule which will The images taken by the camera that are stored on
be used to take photos at regular intervals of the the SD card will be used to create a time lapse using
ascent and descent. This will be Raspberry Pis Adobe Premiere Pro. They will then be stabilised and
own camera module (Figure 7), with a fixed focus colour corrected using Adobe After Effects.
and 5-megapixel sensor. The pictures will be used
to create a time-lapse which will be corrected for Power
The Raspberry Pi, for normal use, requires a DC
power input via a standard micro USB. Whilst
programming a wall socket adapter has been used,
but unfortunately there are no wall sockets 30
kilometres up, so battery power is required. Rather
than buying one of the increasingly popular USB
battery chargers, a battery pack was designed
and built (Figure 8), which uses a step down DC
transformer to convert the 9 volts produced by 6 AA
batteries to the 5 volts the Pi needs for a sufficient
duration of time.

Figure 7: The Raspberry Pi camera

stabilisation and exposure in post-processing on the
ground with a video editing suite.
The camera has been coded with a loop that will
repeat for the number of photos wanted. Within the
loop there is a command to take a photo, after which
the programme waits for a set length of time. This
will be the interval between photos.

The Pi camera came with its own set of commands

which, after downloading, can be called in any script
to take video and images. The python-picamera
commands by Dave Jones were used as they best
fitted this projects needs. Raspberry Pis come with
a camera slot attached to their circuit board, so
connecting the camera consisted only of plugging it
in.
Figure 8: Power pack, including trans-
former and batteries.
ROYAL SOCIETY SPECIAL I ISSUE 17 I WWW.YSJOURNAL.COM 23
 PROJECT A Raspberry Pi Weather Balloon
Discussion and Conclusion and coded on individual Pis, the master code can.
This paper is far from a complete description of our be assembled, and coded to run on start-up. This
high altitude balloon flight and its results, but more a will go on a large 64MB SD card, and then in a
summary of our endeavours up to this date, and how super-lightweight Raspberry Pi model A (Figure 9),
we intend to proceed. which has less RAM but also fewer unnecessary
sockets such as the Ethernet port the other models
The biggest issue and the main task that has have. The challenge then will be integrating the
slowed progression is the problem of getting the Pi wiring for the various sensors, resistors and other
to communicate with the laptop. With the ability to components onto a single tiny breadboard (Figure
code and transmit in RTTY, the rest of the project 10), and soldering each join in place. This will involve
can proceed. With all of the components assembled some code editing as different components need
to use the same GPIO pins, but should not be too
difficult. A spreadsheet on the ground will also be
necessary to process the received data and output
Figure 9: The
graphs, including changes in pressure, temperature
256MB Raspberry Pi
and speed. While the GPS data will give a rough idea
model A, with a
of altitude, the pressure sensor will provide much
heat sink mounted
more accurate data, but raw pressure data must be
on the CPU.
converted into an equivalent altitude. The current set-
up is pictured in Figure 11.

The aim is to launch from Exmoor summer 2015

once the weather becomes more appropriate for a
balloon launch. The capsule and the modules within
it will be tested rigorously beforehand to ensure there
is no sudden malfunction at any point.

Resources
Barometer from Adafruit
GPS from Adafruit
Camera from Raspberry Pi
Figure 10: Raspberry Pi breadboard. Radio module from Radiometrix
Radio transceiver from Yaesu
Temperature module from Dallas Semiconductor

Glossary
Baud rate: a measure of how many times per second a
signal changes (or could change).
Breadboard: a board of holes that are connected
underneath in rows, allowing pins to be inserted in
columns to connect to each other.
Import: tells the code to use a certain library
of commands which can be called later in the
programme.
RTTY: RadioTeleTYpe, a communications system.

Acknowledgements
Figure 11: Temperature, pressure and Mr Tom Wilson, Redland Green School, and Professor
GPS sensors on a breadboard connected Philip Wilson, University of Southampton.
to a Raspberry Pi B.

Author
Redland Green School, Bristol

We are a tenacious group of comprehensive schoolstudentsat Redland Green

School Bristol aged 12-17,who have spent 3 years working on an after school
STEM clubproject to send a balloon to the edge of Space. Authors: James
Hinschelwood, Alex Sidiqui, Issaac Wales, Max Williamson

24 WWW.YSJOURNAL.COM I ISSUE 17 I ROYAL SOCIETY SPECIAL