Andy Garcia · freelance dev
FRENES
Back to the blog
Space

Juice's Battery Bottomed Out at 54% in Earth's Shadow. At Jupiter, Sunlight Is 27 Times Weaker.

October 20, 2026 7 min read

On September 27, 2026, at 21:30 CEST, ESA's Jupiter Icy Moons Explorer, Juice, slipped into the edge of Earth's shadow. Over the night it spent three hours in full shadow, the umbra, where Earth hides the whole Sun, and six in partial shadow, the penumbra, where it hides part of it. That is a record for the mission: its eclipse during the 2024 lunar-Earth flyby lasted 32 minutes. For comparison, ESA notes, people under the total solar eclipse of August 2026 got just over two minutes of totality.

Flight controllers at the European Space Operations Centre watched solar-array power fall and battery use rise as telemetry came in through ESA's Cebreros deep space antenna. The battery dropped to 54% before the arrays resumed generating power at 04:14 CEST. The operational limit was 35%. About nine and a half hours later, at 13:45 CEST (11:45 UTC) on September 28, Juice passed 8,640 km above the Indian Ocean.

ESA published the eclipse figures on October 14, in an article titled "Earth today, Jupiter tomorrow." The title is the point. The flyby was a navigation manoeuvre, but the team used it as a dress rehearsal for the conditions Juice will meet at Jupiter, years before it gets there.

I do not fly spacecraft. I am CTO of a LegalTech built on Claude, and I ship iOS apps and a strategy game as an indie developer. The question I bring to a story like this is how the people who run complex systems test their worst hour before it happens. ESA's three articles on the flyby are short, but precise enough to answer it.

The flyby, in a few sentences of physics

Relative to Earth, a spacecraft leaves a flyby as fast as it arrived. What changes is its direction. But Earth itself is moving around the Sun, so a turn relative to Earth becomes a change of both speed and direction relative to the Sun. That is a gravity assist: momentum borrowed from a planet instead of burned from a tank.

ESA says this one bent Juice's path by 20° and added 3.5 km/s. The navigation campaign started on August 17, with six slots reserved for small corrections, including at four weeks, two weeks, one week and three days out, and ESA's preview had flight controllers prepared for a final emergency manoeuvre at 08:00 CEST on the morning of the flyby, if one was needed. ESA's post-flyby report counts a single nudge out of six opportunities; a week before the flyby, Dietz had placed it at minus four weeks.

"We used only a small amount of the propellant reserved for this flyby," said Angela Dietz, Juice's spacecraft operations manager. "This gives us more to use at Jupiter to carry out observations of the planet's icy moons." On a spacecraft whose full tank holds 3,650 kg of propellant, against a dry mass of 2,420 kg, that is the currency that buys observations later.

Why an eclipse near Earth matters at Jupiter

Juice runs on sunlight. Its two cross-shaped wings carry 23,560 solar cells over 85 m². Near Earth, sunlight delivers about 1,360 watts per square metre. At Jupiter, solar cell engineer Carsten Baur said in a 2023 ESA article, it is "more like 50 watts per square metre, like going indoors." 1,360 divided by 50 is about 27, and that is the inverse-square law at work: Jupiter orbits on average 5.2 times farther from the Sun than Earth does, so the same sunlight is spread over 5.2 × 5.2, about 27 times the area. That leaves about 3.7% of the sunlight Juice gets near Earth. ESA's pages round it to "3%" or "25 times weaker," depending on the page.

Take ESA's figures at face value: 85 m² × about 50 W/m² is roughly 4,250 W of sunlight falling on the wings at Jupiter, for about 850 W of electricity, according to ESA's spec sheet. That has to run the spacecraft and its ten instruments. And when Juice passes behind Jupiter or one of its moons, there is no sunlight at all. ESA's spec page gives those eclipses as lasting up to 4.8 hours; the October article says "up to five hours." Juice carries five battery modules for that. At Jupiter, the battery is part of the operating plan, not a backup.

"This was our first opportunity to test those prolonged eclipse conditions in flight," Dietz said.

What the rehearsal measured

ESA's preview, published September 21, had scheduled Earth's shadow from 21:24 CEST on September 27 to 06:03 CEST on September 28, with the spacecraft "working on battery power alone" and the team "prioritising the most important instrument observations." In the October account, Juice entered the penumbra at 21:30 and its arrays produced power again at 04:14: 6 hours 44 minutes from the first dimming to power coming back.

The battery's low point was 54%, against a limit set at 35%: a margin of 19 percentage points. With the spacecraft behaving nominally, ESA wrote, "the result exceeded expectations."

The team had not assumed it would go that well. "Initially we thought we would not have enough power for the payloads," Dietz recalled. "But after we looked into it further, we realised we could keep several instruments active." All ten science instruments contributed to the campaign, though not all at once during the eclipse. ESA's image page dates the start of the JANUS science camera's operating window to 23:00 CEST on September 27, inside that eclipse timeline, and the German Aerospace Center, DLR, says the camera began with long exposures of Earth's night side while Juice was still in the shadow. By 10:40 CEST on October 2, JANUS had taken 740 images of Earth and the Moon.

Constraints forced choices, and project scientist Claire Vallat described the rule: calibrations "were prioritised based on their relevance to preparing the instruments for their work at Jupiter, while also considering whether an opportunity is unique or can be scheduled later in the journey."

The team also gave EAGLE its first in-flight exercise. The software uses Juice's two navigation cameras to track a moon's illuminated edge and sharpen pointing during close flybys; the Earth-Moon geometry stood in for Jupiter and Io. And with Earth blocking the Sun, Juice could point freely, which the heat of its cruise normally restricts.

The line that matters most to me came from Paula Betriu, a Juice science operations engineer working at ESA's European Space and Astronomy Centre (ESAC) for Starion Spain: "The power data we got from this flyby is very important because it will be used for future Jupiter eclipse simulations."

What the rehearsal does not show

ESA gives percentages, not energy. Neither its flyby articles nor its spec sheet state the battery's capacity in watt-hours or its charge when the eclipse began, so from outside, the 54% cannot be turned into a power budget.

The worst case is longer. Three hours of full shadow is less than the 4.8 hours ESA gives for the longest eclipses at Jupiter.

The exit is different. Near Earth, the arrays came back to full sunlight. At Jupiter they come back to sunlight about 27 times weaker, and in 2023 ESA wrote that panels leaving eclipse there can fall to just 30 degrees above absolute zero. Nothing in ESA's account says the Earth eclipse reproduced that cold, or Jupiter's radiation.

And a first exercise of EAGLE is not a validation. On October 14, ESA said its results "will be analysed in the coming months." As of October 20, I found no published result.

So the 54% is not a certificate that Juice will ride out a Jovian eclipse. It is a calibration point for the model that will plan those eclipses.

What I take from it as someone who ships software

An analogy, with its limits: the nearest thing in software is a game day, where a team cuts a dependency on purpose in staging before production does it for them. A staging environment can be rebuilt every afternoon. Juice gets a handful of these chances in an eight-year cruise, which is why Vallat says the journey "provides only a few opportunities to calibrate and validate the instruments."

Here is what I would copy:

  1. Find the cheapest analogue of your worst hour and schedule it like a release. Dietz says the team prepared this flyby for more than a year, and ESA says eclipse and contingency scenarios were rehearsed months in advance.
  2. Write the floor before the test. "Low point 54%, limit 35%" is a result. "It went fine" is not.
  3. Treat the outcome as input to your model, not as a pass. The value of that night is that the Jupiter simulations will run on measured data.
  4. When the window is tight, rank every test by two questions: does it matter for the real thing, and will I get this chance again?
  5. Reserve correction slots, and do not feel obliged to use them. One nudge out of six left more propellant for Jupiter.
  6. Write down what the rehearsal does not reproduce, next to its result, before someone reads the margin as a guarantee.

ESA plans one last Earth flyby in January 2029 and arrival at Jupiter in July 2031. By then the 54% will be a footnote. The simulations it is meant to feed will not.

Sources

A project like this one?

I design and deploy products like this. Let's talk.

Let's talk