Does “Pix or it didn’t happen” apply to traveling to the edge of space on a balloon-lofted solar observatory? Yes, it absolutely does.
The breathtaking views on this page come courtesy of IRIS-2, a compact imaging package that creators [Ramón García], [Miguel Angel Gomez], [David Mayo], and [Aitor Conde] recently decided to release as open source hardware. It rode to the edge of space aboard Sunrise III, a balloon-borne solar observatory designed to study solar magnetic fields and atmospheric plasma flows.
To do that the observatory needed a continual view of the Sun over an extended period, so the platform was launched from northern Sweden during the summer of 2024. It rose to 37 km (23 miles) and stayed aloft in the stratosphere tracking the never-setting Sun for six and a half days before landing safely in Canada.
Strictly speaking, IRIS-2 wasn’t part of the primary mission, at least in terms of gathering solar data. Rather, the 5 kg (11 pound) package was designed to provide engineering data about the platform, along with hella cool video of the flight. To that end, it was fitted with four GoPro cameras controlled by an MPS340 microcontroller. The cameras point in different directions to capture all the important action on the platform, like the main telescope slewing to track the sun, as well as details of the balloon system itself.
The controller was programmed to record 4K video at 30 frames per second during launch and landing, plus fifteen minutes of 120 FPS video during the balloon release. The rest of the time, the cameras took a single frame every two minutes, which resulted in some wonderful time-lapse sequences. The whole thing was powered by 56 AA batteries, and judging by the video below it performed flawlessly during the flight, despite the penetrating stratospheric cold and blistering UV exposure.
Hats off to the IRIS-2 team for this accomplishment. Sure, the videos are a delight, but this is more than just eye candy. Seeing how the observatory and balloon platform performed during flight provides valuable engineering data that will no doubt improve future flights.
There’s gotta be a reason I’m not seeing… But 56 AA cells? That just seems like a ton of weight unless it’s ballast or balancing.
Lithium-Iron (NOT Li-ion or LiFEPO4) is indisputably the best chemistry for this task – high current output, high energy density, operates down to -60 C and below. And conveniently available in AA size as the Energizer Ultimate Lithium series — even available at Home Depot.
It was a tradeoff between mass budget and cost budgett. Lithium batteries required extensive and expensive testing for this mission, and the intrument was from a group of hobbiest, so we went for NiMh and reduced the FPS for the timelapse.
It was 56 NiMH Eneloop AA cells, (8 in parallel, 7 in series) according to github
Yeah, saw that, was composing my comment below while yours got posted.
Oh, and the lithium AA is just 15 g mass (cf 23 g for typical alkalines).
But actually reading the docs, they use Eneloop NiMH. Not the great choice. Heavy (27 g!), and performance drops dramatically below freezing. BUT, they are rechargeable, which is needed for a long-duration solar-powered balloon, and if they can keep the batteries warm in the 24-hour sunlight it might work fine.
It would be interesting to see what drove the decision to use these batteries and dedicated such a large fraction of the craft weight budget to them.
Hi Paul, I was a member of the team who develop it, so please go ahead with any questions you have :).
The balloon was Solar powered but not this instrument. It had to be fully independant from the telescope, otherwise we had to go through a painful and expensive testing campaign. Battery has been sized following the power budget and the mass budget limitations. Lithium was discarded also to avoid extra testing campaign.
The difficulty of the instrument was to make it fully autonomous, that it is why it has an altimeter and accelerometers., it needed to be as less troublesome to the telescope as possible.
NiMh Eneloop batteries are heavy because they “deliver” exactly the energy they promised. For me they are still the best NiMh batteries in the market. We tried several different batteries and they were mostly fake or they degraded pretty fast (ikea and amazon NiMh batteries perform better than expected but they also degrade fast). This instrument was built in 2021 and has gone through two Sunrise flights, the batteries still performed as new. You can see the telemetries in another post in another comment.
Consider that the 56 batteries account for 1.7kg of the instrument, a lithium battery of the same capacity would have been around 1kg, so at the end of the day, for us the decision was easy, clear and fast, based on experience and too many benefits. :-D
Yes, much more and useful information is available on the parent websites. This post is a little terse for context and easy to miss those details.
But do consider the lithium AAs, like Energizer L91. Compared to Eneloops they perform better in the cold and are MUCH lighter: 4.5 Wh per 17 g cell compared to 2.5 Wh per 27 g, a factor of 2.8x energy density: that 1.7 kg pack could be just 0.6 kg.
Yes, they are expensive and single-use, but they make a lot of sense in this application. There’s a reason Vaisala uses them in weather radiosondes — because they work at -60 C.
When they quote “energy” as “3 Watts/h” and specific energy in “W/g” it does raise questions about their engineering decisions though.
Let me add the telemetry of this instrument during the flight printed in a Grafana panel. For whoever like me, loves the graphs :-D
http://bultza.duckdns.org:3000/
The database is my personal server, so please be nice to it. I hope to put it in a more stable way in the near future for future reference.
As someone who loves both graphs and Granana…thank you, this is glorious. 😻
What’s delightful about balloon trips to near-space is the reminder that, while the cutting edge of technology always belongs to a tiny elite, the cutting edges from past decades are still making their way into everyday life. Like, balloons have been a thing for 240 years and the idea’s still bearing fruit, and most technology is much newer than that.
Nice landing! thunk
Any pictures of stars captured in the daytime? (other than the sun)
So this was actually a piggyback mission on the MUCH larger Sunrise III telescope observatory: 3.5 tonnes! The 5 kg mass budget here wouldn’t be that important. That’s a huge balloon.
Now I’m wondering how you stabilize a 1m telescope hanging from a million cubic meters of balloon. Momentum wheels, yes, but how do you desaturate? Precess and torque against the balloon/gravity vector?
A lot of active compensation. Combination of reaction wheels and compensating motors (plus sensors for the control loop), plus additional compensation in the image path. The first Sunrise flight’s construction details are all here:
https://link.springer.com/article/10.1007/s11207-010-9662-9
Thanks. Cool. It desaturates the azimuth momentum wheel by using two of them in series, the bottom one doing the actual fine control and the top one torquing against the balloon itself. Neat. No fancy precessional gymnastics required. I would not have guessed you could get that much “control authority” from the balloon, but it is huge.
The actual balloon envelope is also a lot heavier than you think it is, too: the mass of the balloon itself (the envelope) is scale-wise on the order of the payload itself! Which means that obviously the moment of inertia is just gigantic in comparison, because the payload’s that much bigger and moment of inertia scales by radius squared.
How can they be sure the ballon won’t land on a populated area? Due to its mass it can/could do a lot of damages (and as it probably did to the trees).
NASA CSBF was in charge of the Gondola, solar panels, balloon operations. They decide when to “cut” the rope to start descend, and the bigger risk is that the telescope can land in a lake as the north pole is mostly unpopulated and the payload is lost. When they make it land there is a recovery team already around in an plane to try to catch where it lands just in case the signal dissapears below the water.
“When they make it land there is a recovery team already around”
Well. That’s the plan. Doesn’t always work out that way, since you can’t control the winds. SUNRISE landed in a “civilized” area (albeit awkward to access), but the balloon can drift to areas that will take time for a recovery team to get to because, uh, it’s basically beyond human civilization.
For long-duration balloons (which have to be launched at polar latitudes) like these, the human impact issue is less of a concern (it’s easy to cut in unpopulated areas). But CSBF also launches short-duration balloons at lower latitudes, and there you absolutely have impact risk: they’ll cut the balloon down before it gets to the point where it can either slowly descend or dead drop and impact something.
Hey, no fair! I did it first!
No, seriously, I did: HELIX launched before SUNRISE and we had a camera for float pictures, too.
A GoPro. Seriously. We figured “why not?”
This is tongue in cheek btw, ours was slapped together at the last minute and although the pictures are cool, this is way cooler. It would’ve been awesome if we could’ve gotten something from landing, too (HELIX landed on a glacier on Ellesmere and tumbled over, which would’ve been dramatic!), but since it was slapdash thrown together there was nowhere near enough power.
A second balloon project I’m on also needs a camera to ensure deployment at altitude, but it’s a much tougher task because that project needs absolute RF quiet so we can’t go the cheapo route.
Do you have any of the data available? pictures or how did you connect it? It would be awesome to share some experiences. By the way, even if IRIS was funded by the Max Plank I am pretty sure that IRIS could fly again (with their permission) in other missions in case you are interested in the future :). The instrument is fully autonomous and software reconfigurable for any lenght of the mission, it can be configured to last for several weeks, probably even more than a month.
We didn’t connect it. We literally mounted a GoPro on the payload and said “hope it works!” and then got the images off at the end. It was totally a bonus.
You can ask the grad student at OSU who actually collected the data (literally) for details (Dennis Calderon-Madera).
I’ve posted a few of them around social media stuff for fun. Dunno if anyone’s gone through and sanitized the images for release (i.e. made CSBF happy), pretty sure we’ve used them in a few PR things.
https://cdn.bsky.app/img/feed_fullsize/plain/did:plc:6zi6e6lhpelbxwemu4aq5v3i/bafkreideqtdur7ont4h3s66k5vkn65xujwlrf4usehxnfnjxgouhagpkxq@jpeg
MPS340 MCU? Are you sure it’s not a MSP430 MCU?