From the article: [...] [...] [...] [...] [...] [...] [...] [...]
From the article:
One of our four CMGs experienced a temperature spike in the flywheel bearing. Our Fault Detection, Isolation, and Recovery (FDIR) logic caught it immediately, spun it down, and executed automated recovery actions. But it wouldn't spin back up. Manual recovery attempts followed. Also unsuccessful.
Rushing back into CMG operations without understanding the failure mechanism risked killing the mission entirely, so we turned off the other three and put the satellite in two-axis stabilization using the magnetic torque rods.
[...]
We had a choice. Hack together novel 3-CMG control algorithms as fast as possible and risk losing another, or figure out how to leverage only the torque rods to achieve 3-axis control with sufficient accuracy to navigate the maneuver to VLEO.
We went with the torque rods.
[...]
We were eager to make some amount of progress, so we started imaging on torque rods even though there would be severe limitations: 50+ pixels of smear, large mispointing from the wobble of torque rod control due to earth's magnetic field, and downlink limited to at best two small images per day. The last two constraints meant we were at risk of spending precious downlink capacity on clouds.
[...]
Although we couldn't control attitude accurately, we did still have good attitude knowledge after the fact. AyJay whipped up a clever idea with Claude Code that automated posting weather conditions in Slack for each collection. We analyzed that to determine which images were likely clear, and selected those for downlink.
[...]
And this is where our ground software really showed its teeth. On most missions, “data on the ground” is just the start — turning raw bits into something viewable is a slow chain of handoffs and batch processing. For us, within seconds of the downlink finishing, the image product pipeline was already posting processed snippets into our company Slack. Literally seconds.
[...]
Three days into the excitement, CMG problems started again.
A second CMG began showing the same telemetry signatures we now recognized as warning signs.
What we had learned from the investigation: the allowable temperature specifications of the CMGs were much higher than the true limit, constrained by what the lubricant inside the flywheel could handle. A straightforward fix for the future — an unfortunate corner case to learn about in hindsight.
[...]
Nine months into the mission, we lost contact with Clarity-1.
[...]
We know exactly what to fix. It’s straight forward: operate the CMGs at lower temperature. The system thermal design is already updated in the next build to maximize CMG life going forward.
[...]
Our next VLEO mission will incorporate these learnings and demonstrate new features that enable missions beyond imaging — we’ll share more details soon. In parallel, imaging remains a core focus: we’re continuing to build optical payloads for EO/IR missions as part of a broader VLEO roadmap.
The successes of Clarity-1 reinforced our core conviction: VLEO isn’t just a better orbit for imaging — it’s the next productive orbital layer.
From the article:
[...]
[...]
[...]
[...]
[...]
[...]
[...]
[...]
People from Albedo Space are answering questions about it on Hacker News: https://news.ycombinator.com/item?id=46747119