In 2024, NASA engineers repaired Voyager 1's flight computer without touching it, by moving software away from a memory chip that had stopped working. The remarkable part is not only the distance but the method: first they coaxed the spacecraft into revealing its own memory, then they cut a program into pieces small enough to fit into the gaps that remained.
A spacecraft that could hear but not speak
On November 14, 2023, Voyager 1 stopped sending readable data. The radio signal still arrived, but instead of structured science and engineering information, it carried a repeating pattern of ones and zeros, as if part of the data pipeline were stuck.
That symptom pointed to the Flight Data Subsystem, or FDS: the onboard computer that collects readings from the instruments and engineering sensors, packages them into a data stream, and hands that stream to the telemetry modulation unit for transmission.
One detail made recovery possible. The spacecraft was still receiving and carrying out commands. Think of a phone whose microphone works but whose speaker only produces noise. You can still give instructions; you just can't hear the replies.
Physics slowed every step. At roughly 24 billion kilometers away, a radio signal took about 22.5 hours to travel each way, so a single question-and-answer cycle took close to two days.
Reading a memory you cannot see
You cannot fix what you cannot diagnose, and the broken FDS could not explain itself. On March 1, 2024, the team sent a command NASA described as a "poke," meant to nudge the FDS into running different sequences in its software in case the fault could be routed around.
On March 3, engineers noticed that one portion of the returned signal looked different from the rest. Decoded, it turned out to be a readout of the entire FDS memory, including its code and the variables it was working with.
This was the turning point. With a snapshot in hand, the team could compare it with readouts from before the failure. On April 4, NASA reported the result: a single chip storing part of the FDS memory, including some of its software code, was no longer working. About 3% of the FDS memory was affected.
The cause remains uncertain. NASA suggested the chip may have been hit by an energetic particle from space, or may simply have worn out after more than 46 years.
Moving a program around a dead chip
A failed chip on a spacecraft in interstellar space cannot be replaced. The only option was to stop depending on it by moving the affected code into memory that still worked.
The obstacle was room. No single free region of the FDS memory was large enough to hold the displaced code as one block. So engineers divided it into sections and stored them in different places.
Picture a library where one shelf collapses, taking with it a chapter readers still need. No empty shelf is long enough for the whole chapter, so you split it into bundles of pages and tuck them into gaps around the building. That is only half the job. Every catalogue entry that pointed to the old shelf must now point to the right bundle, and each bundle must end by directing the reader to the next.
The software patch faced the same problem. The relocated sections had to be adjusted so they still worked together as a whole, and every reference elsewhere in the code pointing to the old location had to be updated. One missed reference could send the computer back into the damaged region or into the wrong instructions.
Engineering data first, science second
The team did not move everything at once. The first step, sent on April 18, 2024, relocated only the code that packages engineering data: the spacecraft's reports on its own health and status. Then they waited for the light delay.
On April 20, a reply arrived. For the first time in about five months, Voyager 1 was sending readable engineering data.
The order was deliberate. Engineering telemetry is the instrument panel. It tells controllers whether the spacecraft is healthy and whether a change caused harm. Restoring that view first meant every later patch could be checked against a working baseline, and a mistake would show up instead of disappearing into more silence.
Over the following weeks, engineers moved and adjusted the rest of the affected code, including the routines that package science data. In May, the plasma wave and magnetometer instruments resumed sending science data. The cosmic ray and low-energy charged particle instruments needed extra work. On June 13, 2024, NASA announced that all four of Voyager 1's operating science instruments were returning usable data.
Why one chip mattered so much
Voyager 1 launched in 1977 and crossed into interstellar space in 2012. With its twin, Voyager 2, it is one of only two spacecraft measuring the region beyond the bubble of solar wind that surrounds our Sun directly from inside it. Losing its data would have meant losing on-site readings of the magnetic fields, plasma, and charged particles out there.
The repair carries a lesson beyond spaceflight. Long-lived systems survive not because they never fail, but because they keep a few capabilities when something does break: a working path for commands, a way to inspect internal state, and spare capacity to work around damage, even if it is scattered. The 2024 fix relied on all three.
It also shows how to work with a two-day feedback loop: change one thing, wait, confirm, then change the next. Voyager 1 will not run forever, since its nuclear power source produces a little less electricity each year. Still, the episode shows how far careful diagnosis can extend the life of a machine that no one will ever touch again.
Loading comments…