Instruments for High-Altitude Balloons

For many of us, an important motivation of launching balloons high in the atmosphere is to put instruments up there. Of course, the fact that instruments are heavier than air means they resist attempts to get them in the atmosphere and are prone to falling out of it. So, airborne sensors need to be light enough to be lifted by an affordable balloon and cheap enough to make risk of crash acceptable. I spent a big chunk of the last couple years designing and building a custom instrument with these characteristics in mind; this post describes how I did it.

My particular field is infrasound, sound waves whose pitch is too low for humans to hear. Infrasound waves normally come from very large sources like lightning, volcanoes, nuclear bomb tests, windmills, and even large animals like tigers or elephants. Unlike audible sound, infrasound waves can propagate long distances without dissipating as heat, which makes them useful for monitoring nuclear and volcanic blasts. Putting microphones high in the air could help improve geographic coverage in our otherwise land-based monitoring networks.

Solar balloons offer an affordable and flexible means of putting cheap, lightweight sensors in the sky. However, commercial infrasound instrumentation falls short in those aspects. To be sure, we do have a few high-quality data loggers and sensors to choose from, but none is lightweight or cheap enough to fly on solar balloons. The closest is the Omnirecs DATACUBE-3 (below, right). Including batteries and sensor, it weighs more than 1 kg and can cost a few thousand USD (depending on quantity and options chosen).

My normal research is on volcano acoustics. I’m not normally worrying about flying microphones on balloons, but I faced similar instrument limitations in volcano fieldwork. For example, my most recent project involved a hike with 2000 m of elevation gain to install 17 instruments near the vent of Tungurahua volcano. On a hike like that, the last thing I need is to be weighed down with heavy instruments, or even heavy batteries for the instruments. Plus, the instruments near the vent are definitely in harm’s way, and I don’t want to accidentally blow up an expensive sensor.

jfa-tungurahua

The upper slopes of Tungurahua: loose, rocky, steep, and prone to ballistic fall during eruptions.

The answer was that I needed to design my own instrument. This process had two parts: firmware (programming the microcontroller) and hardware (finding the right components and connecting them together). Fortunately for me, I began this project around 2012-2013 when a revival in hobbyist electronics was in full swing, and had access to components that are powerful but still novice-friendly.

jfa-logos

The various companies and online forums I used to design the instrument. Most of these cater to hobbyists and beginners.

The Arduino system makes firmware way easier than it used to be. Arduino is a combination of a microcontroller (very, very basic computer), user-friendly circuit board, and programming interface that makes it easy for hobbyists to program the microcontroller and connect peripherals to it. It’s perfect for a beginner. In my case, the firmware’s job was to read data from several inputs, do a little processing, and write output data to a micro SD card. Inputs included an analog-digital converter listening to the infrasound microphone, a GPS (for precise timing and location), a temperature sensor, and a battery voltage tracker.

jfa-arduinoide

The Arduino IDE speeds up the process of writing and testing code.

Without the Arduino system to simplify this process, it wouldn’t really be accessible to someone like me. I’m a geophysicist, not an electrical engineer! But it still wasn’t easy, and it took me a couple years of off-and-on work before I really got it working well.

The hardware part of this project began with a bunch of user-friendly breakout boards wired to the Arduino. A breakout board is a little circuit board that includes some principal component (say, a micro-SD socket, or a GPS) and all the various minor components needed to make it work. The user simply then has to make a few connections (i.e., power, ground, and communication wires) between the microcontroller and breakout board to make it work. Breakout boards are important because a bunch of little things–for example, bypass capacitors and level shifters–have to be added for the principal components to work properly. These little things are often obvious to electrical engineers but can be overwhelming to beginners. Platforms like Arduino and Raspberry Pi might be the foundation of hobbyist electronics these days, but breakout boards make it possible to do cool things with those platforms.

One bit of advice to the DIY instrument designer: if you base your design on the Arduino, don’t use an actual Arduino in the final product. Use a bare-bones variant instead. Real Arduinos have a lot of nice features that unfortunately burn a lot of power, and you can reduce power draw by up to a factor of 30 by using a bare-bones knockoff. I used the Diavolino from EvilMadScientist at first, but am now designing my own board with a built-in efficient power supply instead.

jfa-gps        jfa-adc

jfa-arduino

Top left: GPS breakout board. Top right: analog-digital converter breakout board. Bottom: Arduino Uno, the most common Arduino version.

After getting the breakout boards to play nicely with the Arduino, it was time to design a printed circuit board (PCB) to house them. This might have been the most daunting part of this process. Two circuit design programs are common among hobbyists: KiCAD (free and open-source), and the light version of Eagle (free, not open-source, restricted to non-commerical use, but popular). Like a lot of scientific software, these programs take a while to learn.

In each program, the first step is to draw a schematic showing conceptually how components connect to each other. So, for example, if your circuit included a resistor and capacitor, your schematic would include the resistor symbol connected to a capacitor symbol, but no information about the size, shape, or position of the components.

jfa-kicad_schematic

Schematic editor in KiCAD

After drawing the schematic, the user decides how to lay out the circuit in physical space–specifically, where the components and electrical connections are located on the top and bottoms surfaces of the circuit board. The first step is choosing a physical footprint for each component. Then, a PCB editor window helps the user place component footprints on physical locations on the board and draw wires connecting them. The files created in this process can be sent to a manufacturer (I use Advanced Circuits, based in Aurora, Colorado, USA).

jfa-kicad_pcb

PCB Editor in KiCAD

The final product is called the Gem infrasound logger (named for the “Gem State” Idaho, where it was developed). Parts for it cost about $250 and it takes a few hours to assemble. This is about an order of magnitude cheaper than the cheapest commercial alternative. On the other hand, I put an uncountable number of hours (hundreds, at least) into designing and revising the Gem. On the other other hand, I learned some valuable skills in the process, and my next electronics project (a power-efficient Arduino board) was much easier as a result.

And of course, it’s turned out to be useful in applications beyond volcanoes. Danny Bowman (of Sandia National Labs, and this blog) has launched them on balloons. Tim Ronan (grad student at UNC-Chapel Hill) used them in his research studying river rapids. (Turns out that being lightweight is important when microphones have to be installed by kayak through class-IV whitewater!) Danny and Tim both helped me a lot by testing Gem versions that were still in development, for which I am grateful.

Was it worth it? As a grad student, my time is cheap and half my job is to learn new skills, so spending a lot of it learning how to do this was a good investment. If I was a professional scientist whose labor was more expensive, it might not be such a good decision: better to pay an engineer to do it, even if the product costs more to develop.

In any case, the final product justifies the effort, regardless of who did it. I now have a cheap logger that I can put high on volcanoes and on balloons. It’s been used in several projects around the world now, and it will remain a key part of my infrasound research.

jfa-solarballoon

A solar balloon carrying a Gem lifting off in Albuquerque, NM, USA. Photo by Danny Bowman.

jfa-sites

A subset of the many sites where the Gem has been tested.

Stratospheric Solar Balloon Flight

It took three years and several false starts, but we finally got a solar balloon with a video camera on board to take off successfully.  Not only that, but the balloon reached an elevation of over 22,000 meters (72,000 feet), well into the stratosphere!  Then, it suffered what Elon Musk calls a “rapid unscheduled disassembly,” sending the payload into a 22 kilometre free fall that ended in the muddy banks of a cow pond.  Despite this, the photos, video, and GPS track were all recovered.  Here’s the flight video:

The balloon envelope consisted of a 19 foot diameter sphere of 0.31 mil clear plastic paint drop cloth.  Since the drop cloth comes in 12 x 400 foot sheets, we constructed the envelope from 5 gores that were 12 feet wide at the equator.  We attached the gores together using clear shipping tape and darkened the interior of the balloon with black paint pigment.  Each gore seam had a string attached to the bottom with black duct tape.  The opening at the bottom of the balloon is about 6 feet across, allowing two people to inflate it by hand (check out the full inflation and launch video).

The solar balloon in flight just after launch.  Image credit: Mary Lide Parker

The solar balloon in flight just after launch. Image credit: Mary Lide Parker

The payload consisted of a SPOT satellite tracker for recovery, an Arduino Uno with Adafruit High Altitude GPS Shield for trajectory determination, and a Raspberry Pi with camera module for video and stills.  The system was powered using a lithium battery pack meant for recharging cell phones (10 ampere hours).  All this was contained in a Tupperware box with a small hole to admit the camera lens.  We attached it to the balloon using four strings taped onto the box with white duct tape.  The four strings led to a fishing swivel to keep the payload from spinning too much.  We clipped it to the strings on the envelope using a black carabiner. Total payload weight was 800 grams (1.75 lbs).

payload

Payload with the lid off, showing flight instrumentation (left) and payload just before launch (right).

We kept an eye on the wind profiles for about 5 months, since most of the time winds in the troposphere blow out to sea.  Finally, we waterproofed the payload and decided to risk a flight even if some winds were going east.  On May 29, the winds in the troposphere were pretty low (max 10 m/s or so) and with varying azimuth.  The stratosphere had a steady breeze going west, so we figured if we made it that high, we’d head back over land.  However, we had a much faster ascent rate and reached a much higher altitude than we anticipated, so we ended up not flying very far from the launch site.  You can download the trajectory data in text format here or Google Earth KML here.

trajectory

Ground flight path (left), launch site is the origin. Altitude versus time (centre), local time was GMT – 4. Ascent rate versus time (right), local time was GMT – 4. I obtained the ascent rate by calculating a 1 minute moving average and dividing elevation by time.

The photos and video were very good quality considering that we were using the Raspberry Pi camera module (not the world’s most advanced camera).  The troposphere was pretty misty, and it seems like we even passed through a haze layer on the way up.  However, it could be that the lens fogged up temporarily.  Once we entered the stratosphere, the pictures are much better:

00670

The view from 22 kilometres in the sky.

You can watch a slide show of all the photos here, and all the video clips stitched together here.

Just as we approached neutral buoyancy, the payload unexpectedly separated from the envelope and fell back down to Earth.  We kept GPS tracking until about 18 kilometres elevation, and as far as I can tell the box was falling at about 320 km/hr (200 mph).  The Arduino kept track of time even after losing GPS fix, continuing to record until the moment of impact.  Thus, we know that the payload fell the remaining 18 km in about 12 minutes.  The impact speed was probably less than 100 km/hr (60 mph).  The Tupperware payload box was cracked, and everything except the SPOT tracker stopped working.  Had we landed 10 centimetres or so west, we would have splashed down in a pond.  Luckily, we hit the mud on the pond’s edge (and missed the cows that were in the area).

CGasLNHUQAAnDHG

The payload box at the impact site in Snow Camp, North Carolina.

We consider this flight mostly successful.  Our main objectives were to launch, recover, and extract data from an instrument package lifted with a solar balloon.  Furthermore, we wanted to inflate the solar balloon by hand.  Both of these objectives were reached. Our secondary goals were to fly until sunset, not land in the ocean, and make it to the stratosphere.  The flight was only about 2.5 hours, so we did not fly all day as we hoped.  However, we made it well into the stratosphere and were never in danger of ending up in the ocean.

We had a slightly tense launch when the payload got snagged on the eaves of a nearby building, but the balloon built up enough lift to detach itself in about 20 seconds.  The SPOT tracker did not record any positions during flight, so we did not know where the balloon was and were not even sure if the tracker was working.  In fact, it did not record positions until about an hour and a half after impact. The unexpected flight termination was upsetting as we did not anticipate having the payload detach from the envelope.  We assumed that either the envelope would rupture due to sun-induced heating at altitude or that it would deflate at sunset.  In either case, the payload would have had a large plastic streamer to slow it down to safe velocities.  Instead, it appears that the black duct tape we used to attach the payload strings to the envelope got too hot in the intense sunlight at 22 km.  This caused the payload to come loose from the balloon.  The lesson we learned from this is to never use dark coloured tape if there’s a chance the flight system will make it to extreme elevations.

A flight aboard the NASA High Altitude Student Platform (HASP)

The NASA High Altitude Student Payload (HASP) project provides a spot on a high altitude balloon payload for undergraduate and graduate students. When I heard about this last year, I gathered a team together, we applied and were accepted into the program. Our project: launch infrasound microphones into the stratosphere. Infrasound (sound at frequencies below audio range) is usually measured at the Earth’s surface, but we know it propagates hundreds of kilometers upward into the atmosphere. Our goal is to measure these sound waves as they cross the stratosphere.

The HASP project was definitely a commitment. As the team leader, I was required to write a monthly status report letting the HASP project leaders know what I was up to. I had to build my payload box under strict power draw, weight, and size limits. I also had to learn electronics from the ground up. Thankfully, another member of our team had lots of experience in electronics, so it wasn’t so bad.

Our Omnirecs DataCube logger installed in the payload box.

Our Omnirecs DataCube logger installed in the payload box.

A few weeks ago, I traveled to Palestine, Texas to bring my payload to the Columbia Scientific Balloon Facility (CSBF). There, our payload was subjected to extreme temperatures (ranging from -50 to 50 Celsius) and pressures (sea level to stratospheric).  We passed the test, recording the 8 Hz signal from the vacuum pump clearly even when the pressure was around 5% of sea level.  This was an important milestone: not only did it clear us for flight, but it also showed that our differential pressure microphones (constructed by Dr. Jeff Johnson at Boise State University) would operate in a near vacuum, something they were not designed to do.

Our payload about to face the thermal/vacuum test.

Our payload (the white box with the UNC logo) about to face the thermal/vacuum test.

From Texas, I traveled to New Mexico to launch our payload into the stratosphere.  I ended up staying in my home town, about 2.5 hours from the launch site at (a CSBF facility in Ft. Sumner, NM).  Needless to say, I spent a lot of time on the road!  I had to drive there and back three times: once to put my microphones on the flight ladder (see below), another time to make sure everything worked during the “hang test” (a dry run for launch), and finally for the big day itself – the flight.

CSBF staff mounting the infrasound microphones on the flight ladder.

CSBF staff mounting the infrasound microphones on the flight ladder.

Like any balloon flight, this one depended on the weather.  This time, the news was not good.  One group was ahead of HASP, and they had dibs on each launch window.  They tried twice, and were not able to fly both times.  I had to fly home on Saturday, so I showed CSBF and HASP personnel how to set up my payload, and I resigned myself to not seeing the balloon fly.

But as luck would have it, the previous team decided to wait, and a launch window opened Saturday morning, the day I was scheduled to fly home.  Since the flight was early in the morning and my plane ticket was for early afternoon, I decided I was going to go see the launch.  I drove out, arrived in Ft. Sumner at about 10 PM, slept in the back of the car for a few hours, then got up at 2:45 AM Saturday morning to start getting ready.

It's about 3:45 AM, and CSBF staff secure the microphones after I powered them up about 15 minutes before.  The wheel on "Big Bill" (the launch vehicle), is taller than I am.

It’s about 3:45 AM, and CSBF staff secure the microphones after I power them up. The wheel on “Big Bill” (the payload vehicle), is taller than I am.

The launch was touch and go the whole time – we had to wait for the winds to all blow in the same direction for the first 1000 ft in order to start the inflation process.  As luck would have it, they did straighten out, and the call was given to roll out and start inflating the balloon.

The balloon reaches full inflation about a half hour after sunrise.

The balloon reaches full inflation about a half hour after sunrise.

The launch was spectacular.  The balloon was released and drifted into the air.  Big Bill started driving in the direction that the balloon was going, and just when it was overhead, the payload was released.  The entire structure (800 ft high!) was now in free flight.  It seemed to climb slowly, but that was an illusion – when the balloon was 12,000 ft above the ground it still seemed close enough to touch.

The radio crackles and says "it's your balloon."  With that, the 3 million cubic ft envelope is released and begins climbing into the air.

The radio crackles and says “it’s your balloon.” With that, the 3 million cubic ft envelope is released and begins climbing into the air.

Here we go!

Here we go!

 

The balloon flew for about 8 hours, and was terminated over northeast Arizona.  Once the recovery team picks it up and ships us our data logger, we can find out what we heard up there.

A big thanks to the Louisiana State team for running HASP, and all the great people at CSBF who made it all happen!