Yesterday, during the total solar eclipse, I launched a high altitude balloon. This balloon took years of effort across several volunteers. The goal was to capture video and imagery from above the clouds. This data would be streamed to the ground, and then relayed to the Internet so that anyone could watch in real-time. My contribution was the high speed downlink between the payload and the ground. I've written about it before, but the summary is that almost all of the hardware/software was developed by me, and it uses the 70cm ham radio band.
We launched from Florenceville, New Brunswick. It's right in the centre of the path of totality, meaning the wind can potentially blow the balloon quite far before it becomes an issue.
Initial preparations
The day before the eclipse, I drove up to Florenceville to do a last minute end-to-end test of the system. David, the main guy behind the project, lives there. We did the tests at his house. I did find a pretty small bug in my ground station software, which I fixed in about half an hour. Had I not been able to fix it, I would have been comfortable leaving it as-is. Basically, the ground station software displays the last two received images. In some cases, the latest image would overwrite both displayed images. Like I said, not too big a deal, but glad I was able to find it and sort it out.
There were also a few bugs which I knew about, but didn't fix:
During one test, the ground station stopped decoding packets properly after about an hour and a half. After restarting the software, the problem went away. I was never able to reproduce this. I wasn't too concerned though - I could always restart the software if it happened again, since it was only an issue on the ground station. I think it may be related to the SDRPlay RSP1a I was using. As usual, I will say I regret buying the RSP1a. The driver is fairly buggy and it's also proprietary. I think the much cheaper RTL-SDR would have done a much better job. I wish I had tested it on a previous launch, but without the confirmation I didn't want to change out such a critical component.
Another bug I've found is that the payload stops sending data properly after about 16 hours. I've observed as low as 9 hours and as high as 25 hours. I've traced it to a Gstreamer bug, which I reported here. Essentially, the video stream starts using more and more bandwidth until there's none left over for anything else. I didn't notice this until after our final test launch, and I definitely wasn't going to risk changing the payload software. Knowing that the software only needs to work for 3 or 4 hours max, I decided to leave this in as well.
Satisfied with the test, I left and went home.
Eclipse Day
With communications so essential, I was worried something might go wrong. Off the top of my head, here are a few potential issues I thought of, in no particular order:
Leaving a critical component at home
Car problems of some sort
Traffic from the eclipse
Some kind of interference causing problems to the ground station
I was able to mitigate some of these concerns. I completely assembled the ground station in my car before leaving to make sure I didn't forget anything. I was especially worried about leaving a pigtail or something at home, rendering my station useless. I also turned off the CATS transceiver in my car. It runs on the same 70cm band as the payload, so I figured I might as well be safe.
With launch scheduled for 3:45pm, I drove up to Florenceville to get there at noon. I had absolutely no car problems (actually, I've never broken down or anything before) and traffic was completely fine. Setup was relatively mundane - this was my fifth time setting up. One major difference was the amount of people around. It kept growing until eclipse time - more and more people around the perimeter of our launch site. There was a ton of different people from the media there, as well as the premier of New Brunswick (Blaine Higgs) and Chris Hadfield. There must have been many hundreds of people at least.
There was quite a lot of wind, but we were able to launch without a hitch (if a bit delayed). Telemetry worked almost perfectly. One issue was that we put aluminum tape on the payload. The hope was to cut down on static buildup, which we kind of (but not really) thought might have been causing other issues. Unfortunately, the tape caused about 20dB-30dB of attenuation. It was possible to see - as the payload spun around, the signal would rise and fall. For context, 30dB is the equivalent between a 1W signal and a 1mW signal. It's also the equivalent between receiving a signal at 50 km and 1600 km. This was very visible on the stream - some frames would come through corrupted, and large chunks of images would be missing entirely. I tried to compensate by pointing my antenna as accurately as possible, but unfortunately it did mean we were losing data by the time of totality.
Aside from that, another small issue is the amount of rotation of the payload. Most people who have watched the stream have brought it up with me - the payload spins around a lot, and it can be quite nauseating. Our payload is very long and skinny vertically, which minimizes its moment of inertia. I think it would have been better to make it long horizontally. Ideally, the mass would be out near the edges.
Onboard we had a camera that pointed at the sun, in spite of the amount of rotation of the payload. I didn't contribute to it, so I don't want to comment on it too much. I will say, I thought it worked very well and I was extremely impressed by the images that came back from it.
Overall, I'm very happy with how it went. The recording of the stream is still available online.
Totality
Aside from the balloon, I got to experience totality. This was my first time seeing a total solar eclipse. It was amazing. I'm extremely happy I got to see it, and I definitely want to see more in the future. The pictures online don't do it justice. Aside from the awesome black circle in the sky, the whole environment changes. It gets dark, but not red like at sunset. Everything just looks... weird? It's hard to describe.
Three minutes of totality sounds like a long time, but I'm amazed by how fast it passed. One member of the group set a timer at the start of totality. When it went off I thought he set it wrong - I thought it had been 15 seconds; maybe 30 max. But no, somehow over 3 minutes had passed without me noticing, and shortly after, the sun came back into view.
Future work
This was an awesome project to be a part of for the last year and a half. There's still more to do - documentation (like this post!), post-mortems, etc. We're going to continue having our weekly meetings for at least a little bit, it sounds like.
I also really like doing balloon launches. I want more of that, I think. I have some refinements I'd like to make to my communications system, and now that I'm collecting radiosondes, I'd like to relaunch them as well. I've got a tank of hydrogen on order and I guess I'll need to assemble a team of people who also want to do balloon launches. High altitude balloon clubs aren't that rare - maybe I'll be starting one in Fredericton? Who knows!