< Back

TELL Rocket Team, May 2018

The Swiss Space Initiative

Taking the Initiative

In October 2017, while checking out the new Student Project House at ETH Zürich, I stumbled upon a very interesting bulletin. Someone was recruiting a team of students to build a rocket for the Spaceport America Cup. Never would I have thought to have the chance to work on a rocket as a student. I directly applied and met Oliver who was leading the project, based on his experience with a similar team at EPFL. Since my experience was with programming, I joined the avionics team together with Raphael and Alessandro. When I learned that the avionics were not meant to interact with the flight but only record data, it was clear for me that I'd want to go further than that. So I found Gianni, a mechanical engineer with the same goal. The expectation of the competition was to reach a specific altitude as exactly as possible, so we pitched the simplest possible control system: an overshooting rocket with air brakes.

To get a working control system we'd have to test it as much as possible, so Gianni and I instantly started working on our first rocket, Mestral. With much support from Jürg, an experienced Swiss model rocket builder, we just managed to finish the rocket for a launch on 18th November in Kaltbrunn. Mestral was built completely from off the shelf parts, with our own addition simply being a Raspberry Pi with an altimeter and IMU. The launch went well and we were able to reconstruct the trajectory from our IMU data. We observed that the altimeter was very sensitive to pressure changes due to flow outside of the rocket. Because of that and since the flight only lasted a few seconds, we decided to use the IMU integration as our main control input source.

For our next iteration, Gianni came up with an air brake design for Mestral. It would include cutting the rocket into two parts, an upper and a lower body with a support structure in between. That support structure would also hold the air brakes. We also added electrically controlled black powder charges to the upper stage for parachute ejection. On the next flight on February 24th 2018, we tested the new parachute ejection system with two black powder charges. One would be controlled by an off-the-shelf altimeter and the other one by our Raspberry Pi. For some reason, the off-the-shelf altimeter didn't fire, but luckily our own software worked and Mestral was recovered safely.

In parallel to the Mestral tests, the development of TELL, our main rocket, continued. Since it would include components from many other teams and both the structure and the electronics PCB was built from scratch, it would be infeasible to get an early prototype. Our control team was soon joined by Serj from the Structures team, who would design the air brake system, and Thomas, a student with a rich background in control theory. Thomas suggested an optimal control algorithm that'd allow us to reduce the real-time control logic to a look up table. This would make the algorithm easy to run, even on our rather slow microcontroller. The main downside of that algorithm is that it relies on an accurate model of the rocket's dynamics. Since we didn't have any good cfd simulations and no test flights of the actual rocket, it was hard to predict the aerodynamic properties. We surprisingly got the chance to test our rocket in a wind tunnel, generously provided to us by Sauber. Unfortunately, the wind tunnel topped out at 80 m/s, which is much slower than the rocket actually flies.

Mestral at its first launch
First iteration of Air Brakes
Visualizing flow lines in the wind tunnel
Mounting TELL: David, Serj, Amir and Anna

The Spaceport America Cup

On Saturday evening, the 16th of June, we arrived in El Paso, Texas. A small team that had already arrived earlier had organized cars and picked us up for the drive to Las Cruces. The next day, we quickly got to reassembling the rocket and testing the functionality. For me, that also meant implementing the functionality that we didn't really get done before. The hotel had an elevator that I could use to calibrate the IMU integration. We also got to enjoy some of the great beef steaks that Las Cruces had to offer.

On Tuesday, we got up early, packed up TELL and went to show it off to the other teams at the conference center. This was also the day Gianni and I would be presenting the control algorithm. So the two of us spent a significant amount of time outside, preparing our slides. Around noon, we did our presentation which was well received by the audience and we even got some interesting questions. It would only have been better if Thomas, who had kickstarted the entire theoretical approach, could have been with us. Afterwards I went straight to bed.

The next day, we once again got up around 5 am. This time we were going to the assembly area in the New Mexico desert about 80km north of Las Cruces. Once there, we had about 24 hours to have the rocket ready for launch. While the plan was that we'd only have to unpack our rocket and reassemble it, none of that worked for the avionics team. Firstly, the IMU integration that I had just tested so extensively in the hotel didn't work anymore. It turned out, that the IMU measurements were suddenly all wrong, meaning the rocket would sense itself at a 30 degree angle when it was pointed straight. It might have been the transport or the environment - the 43° C heat and dusty wind certainly didn't help us either. So I spent most of my time during the day adding a compensation for the sensor bias. Once it turned dark, at least the weather became better for working. We started assembling the avionics and everything seemed to work, until we did some basic shock testing. That means hitting the nosecone on the table. We weren't exactly sure what was happening, until we realized that the entire PCB was losing power every time we hit it on the table. The components, especially the power connector, hadn't been soldered well and were all starting to come loose. Alessandro used this chance to show off his soldering skills and resoldered all of the critical components using a gas-powered soldering iron. In a last, unfortunate mishap, we destroyed the camera that was supposed to film the launch from the rocket. It was attached by a cable to the avionics, which had to be disconnected before removing the avionics. We forgot to do that and ripped the camera apart. Although technically there was a spare part for everything, in the dark of the night we were unable to find it. At that point we had about two hours left until it was time to get up for the launch, so I tried to get some sleep but there wasn't really any space left in the RV.

Without having slept really, we started assembling the rocket for launch a few hours later. Unfortunately there were some delays that pushed our launch back to around noon. Then, we finally got our moment of gratification where we could sit back and watch the launch. Except that we suddenly got concerning reports from the EPFL team that their air brakes might have caused the nosecone to prematurely eject and impale their lower stage into the side. Since our rocket was already on the rail, there was nothing we could do but watch now. When it came to the launch, TELL cleared the launch rail nominally but then the engine started misbehaving. Two seconds into flight, the solid rocket motor exploded, making short work of what we had built over the past year. At this point, I wasn't even phased by the accident. For all I knew, all of the systems we had built, worked. Out of all the parts that could fail, it was the one we bought off the shelf. In the afternoon, much of the engineering team found themselves a spot in the shadow to sleep off some of the stress. To add to the bad luck, we were all so focused on the rocket that we hadn't really brought enough food for everyone. Even though the plan was to camp out in the desert, that evening I returned with some others to the hotel in Las Cruces to recover.

On Saturday, the launch site was cleaned up and everyone returned to the city. There, the prices were to be announced. Obviously we couldn't have won the general price, since our flight performance was non-existent. Still, the panel of judges recognized our hard work and rewarded us with the Charles Hoult Award for Modeling and Simulations. As a bonus, someone from our team also managed to negotiate ourselves an exclusive tour of the Spaceport itself. So the next morning, we drove back out to the launch site and had a look at their facilities for full-scale launches. In the afternoon, we made our way southeast to the White Sands National Park. While the area around the launch site was hot, dry and dusty, it was still a thriving ecosystem in comparison to the White Sands desert where there was barely anything living. Amir, Peter, Raphael, Christian and I continued from this point on our own. We had the plan to make our way to Santa Fe that night. Coming from our small country, we clearly underestimated the distances. We couldn't even switch out drivers due to the car lending agreement, so Raphael heroically took the wheel for the entire night. Along the route we saw some interesting places, including a small village called Corona featuring a phone booth with an alien inside. We arrived at the motel way too late, but were still able to get our room and spend the night.

Now we had Tuesday as a well-deserved day to relax and have some fun in Santa Fe. For me the highlight was an exhibition on the Manhatten program at the New Mexico history museum. The next day, we would make our way to Albuquerque. Since the drive was much shorter than the last, we planned a stop at the Tent Rocks monument, featuring a very tight canyon and some nice volcanic rock formations. On Thursday we got a proper taste of American culture by visiting a large gun store in Albuquerque. In the evening, we tried to go stargazing since I missed my chance to do that in the desert. Unfortunately the moon made it impossible to see much.

For Friday, our last full day in New Mexico, I still wanted to go on a hike. Amir and I were able to meet up with Alex and Robert who had come to Albuquerque independently. I thought it'd be easy to find a place to hike some nearby mountains, but there were some problems. The first road we took, was in much worse shape than I'd have ever expected a road to be, so we had to turn our car around. That explained why we were seeing so many trucks on the road. Then, we realized that most roads were part of private property, and if it wasn't, somebody was living in their RV there. In the end we still found a place to park our cars and a little forest to explore. Unfortunately we ended up running into rain, but that was completely worth it for the experience.

Since we were unable to actually test our air brake system, on 6th October 2018 we flew Mestral again. Now with functioning air brakes and a level 2 motor that'd allow us to reach nearly 900 meters. We aimed for 800 meters and our air brakes managed to get the rocket to an apogee of 808 meters. Not quite what we wanted, but only a 1% error! Thanks to Thomas' initiative, we were then able to publish these results at the 8th European Conference for Aeronautics and Space Sciences. Click here to view the paper.

Me inserting the ignition charge into the motor