Balloon Launch 2 – Achievements and Plans

The High Altitude Balloon Team’s final project objective is to design and build a balloon payload that holds and stabilises a ground-facing radar for surveying purposes. To achieve this, we have to collect abundant amounts of data to help us design a payload enclosure and stabilisation system that will function in the intended operating environment.

Launch 2 of the Balloon Team yielded average quality data. It showed us some trends and a portion of the general behaviour of atmospheric conditions and our payload’s response. Unfortunately, our sampling rate was 2Hz so the highest frequency we could detect was 1Hz. Higher frequency oscillations—which is predominant according to video footage—could not be registered. Despite the limitations of the data collected, we still analysed it as it captured lower frequency oscillations and provided good practise for future data analyses. Future launches will sample data at a frequency of 200Hz.

Analysis of Data

A short-time Fourier transform was performed on the acceleration data in the x-axis (the x-y plane is the horizontal plane) with a time window of 128 seconds (248 data points) and a 64 second overlap between windows. This was then plotted as a 3D surface with logarithmic elevation that is also simultaneously marked out by colour. The resulting plot is shown below.

Graph representing data logged by the high altitude balloon sensor payload


As can be seen from above, there is a slight peak at around t = 9000 samples corresponding to the ratio f/F = 0.19 where F is our sampling frequency of 2Hz. This means at around 16km altitude, there were stronger-than-usual oscillations of 2*0.19 = 0.38Hz. An oscillation frequency of 0.38Hz implies one complete swaying motion every 2.63 seconds, which is comparatively slow and gentle.

At around sample no. 13000, which was taken at 23 km altitude, the amplitude of the signals becomes much higher over the entire frequency spectrum. This is due to the balloon burst and the subsequent chaotic descent during which the payload shook around violently. This was caused by high falling velocity and non-uniform opening of the parachute. However, our mission objective requires stabilisation only while the payload is suspended from the balloon. After the balloon stops providing lift, we desire only that it falls at or slower than 12m/s. Everything else is redundant.

What We Did Right

Although we made mistakes in the lead-up to, during, and after Launch 2, we also did quite a few things right. We will be replicating what we did right for future launches. Telemetry is one area that we believe was done mostly well. The data we got back is a bit iffy but it generally shows a trajectory that is within expectations. It’s also interesting to superimpose altitude against acceleration so we can see the different forces acting on the payload throughout its journey, as below.

A graph describing the acceleration data relative to the balloon's altitude

As is apparent from above, the point where the payload begins to drop in altitude corresponds to massive increases in forces on the payload. This immediately tells us the balloon burst was a chaotic event (confirmed by video). The z-acceleration also switches from oscillating around positive 1g to negative 1g making it clear the payload flipped upside down. This was by design as our payload mounted the parachute at the bottom of the payload enclosure. About halfway through its descent, the payload’s shaking significantly reduced. We believe this is due to a rapid transition from low-density to high-density air, which stabilised the descent by fully opening the parachute.

Although the ascent data was generally good, data of the descent phase is highly suspect due to extremely large errors between each sampling point and very sporadic data logging. Having few data points reduces the reliability of whatever data we’re able to retrieve. We hope to improve our telemetry data by changing to a more reliable and accurate data logger.

Another thing we did quite well is enclosure insulation. We had both external and internal thermometers on our payload. At the most extreme, they measured over 100 degrees difference between the external and internal temperatures. As electronics and batteries begin to fail at the sorts of external temperatures we measured at high altitude (-67 degrees Celsius), we were pleased that our insulation kept the internal temperature at a comfortable 30-40 degrees C. We will definitely be using similar insulation on future launches.

The Next Balloon Launch

The data we’d collected from Launch 2 is not terribly useful but we nonetheless have a mission plan for the next launch. Passive stabilisation is the area of focus for the next launch. Passive stabilisation means there is no external intervention to stabilise the system, i.e. no motors or moving parts to steer the system back into the desired position. A simple example of passive stabilisation is a pendulum; when the mass is moved from its equilibrium position, it will want to move back to it after you let go, and if you let it swing for a while, it will eventually come to rest at its equilibrium position. This is an example of passive stabilisation as the system stabilised itself without needing external intervention.

Passive stabilisation is an important part of the overall stabilisation scheme as it simplifies the active stabilisation system’s job by damping oscillatory motion of all frequencies. Two methods of passive stabilisation are being explored for the next launch:

  1. The first involves extending point masses outward and downward from the bottom four corners of the payload enclosure. This increases the moment of inertia in the horizontal plane and about the vertical axes making it harder for the wind to sway and rotate the payload.
  2. The second is a powered gyroscope-type system contained inside the payload enclosure with the device-to-be-stabilised’s motion coupled to the rotor’s, thus matching the latter’s attitude. In theory, the rotor’s attitude—and therefore, our device’s—should remain constant as long as it maintains high angular momentum. A motor will continuously be driving the rotor so it spins throughout the balloon’s flight.


Overall, we do not consider Launch 2 a 100% success. However, our team is all the richer for having done it. We’ve been taught a valuable lesson on good data collection as our analyses were less than ideal because of flawed collection. In spite of that and regardless of how useful the analyses were, simply doing them is fantastic preparation for analysing the good data that we know we’ll get from our next launch.

We also learnt a lot about mission preparation and execution. Launch 2 was a bit rushed; there were a few hiccups during the balloon inflation and our schedule slipped multiple times. We are now aware of the need to be better prepared not just for the planned sequence of events but also likely contingencies that may arise. This might not eliminate the hiccups on our next launch but it’ll certainly allow us to deal with them more efficiently.

On a more casual note, the trip was just plain fun. We had lively conversations with each other and even had time for board games. Although I may have given you the impression that a balloon launch is stressful and scrupulous, the reality is it was actually quite relaxing. It was like playtime except instead of children’s toys, we played with science-y gadgets and a high-altitude balloon. I, along with everyone on the team, look forward immensely to our next balloon launch where more fun awaits!