A CubeSat that flies itself to catch solar storms, and how Earth out-listened a fading Voyager

01 A launch contract, and a first

ESA's technology directorate spent the week of 16 July 2026 closing the piece of the plan that had been missing, a ride. The HENON consortium signed Arianespace to carry the spacecraft as a secondary payload on the Ariane 6 that will also loft PLATO, ESA's exoplanet-hunting telescope. [1][4] The agency lists the target as early 2027, though PLATO's own schedule has been quoted as late 2026, so the exact flight date is still soft. [1][6] HENON is built by the Turin firm Argotec, leading a consortium drawn from Italy, the United Kingdom, Finland and the Czech Republic. [2][4]

What ESA claims is narrower than "first CubeSat in deep space," and more interesting for it. HENON is meant to be the first CubeSat to travel into deep space on its own, talk directly to Earth, and steer itself to a destination without riding inside or alongside a larger parent craft. [1] Earlier deep-space CubeSats were deployed by host missions and largely coasted or made small corrections; this one is expected to navigate and propel itself the whole way. [1][2]

02 The machine, in numbers

The airframe is a 12U XL CubeSat, a box a little larger than a carry-on bag, at roughly 30 kilograms fuelled. [2][3] Getting from the launch drop-off near the Sun–Earth L2 direction to the operational orbit is the job of a single radiofrequency ion thruster, an engine that ionises propellant and accelerates it with an oscillating electric field. [2][3] It produces about 1.7 millinewtons of thrust, roughly the weight of a housefly, at a specific impulse near 3600 seconds, a measure of how much push each unit of propellant buys, and draws on the order of 80 to 130 watts in cruise. [3] Low thrust demands patience, and the mission analysis puts the spiral out at about 389 days, with the engine shut down weekly so the ground can track the spacecraft. [3] The total velocity change is near 1.3 km/s, bought with roughly 1.55 kilograms of propellant. [3]

The science payload is three small instruments. MAGIC, a magnetometer from Imperial College London, weighs 23 grams and uses anisotropic magnetoresistive sensors rather than the heavier fluxgate design, about ten times lighter for the same job. [5][2] REPE, from the University of Turku, counts energetic electrons and protons; FCA, a Faraday cup from Charles University in Prague, measures the bulk solar wind. [2] Together they sample the plasma and its embedded magnetic field in a region no spacecraft has occupied. [2][3]

03 Why upstream is the whole game

Whether a coronal mass ejection wrecks a power grid depends less on how much plasma arrives than on which way its magnetic field points. The damaging case is a southward field, Bz, that reconnects with Earth's own northward field and pours energy into the magnetosphere. [3][5] That orientation cannot be read from a telescope or inferred reliably from a model; it has to be measured by an instrument sitting inside the cloud as it passes. [5] So the forecast is only as early as the most distant sensor in the solar wind.

Warning time is just upstream distance divided by the storm's speed. The L1 monitors, ACE and DSCOVR, sit about 1.5 million kilometres sunward of Earth, which for a fast CME travelling one to two thousand kilometres per second is fifteen to twenty minutes of notice. [5][3] Push the sensor out to roughly 15 million kilometres and the lead grows by an order of magnitude, to a few hours. [3][5] The hard part is staying there. A point that far up the Sun–Earth line is not gravitationally stable, and holding it would burn propellant a 1.5-kilogram budget does not have. [3] HENON instead flies a distant retrograde orbit, a large loop co-orbital with Earth that stays put for more than a century with almost no station-keeping. [3][7] That orbital-mechanics choice is what lets a 30-kilogram satellite live where a much larger one would otherwise be required. [3][7]

04 What to watch

The near-term marker is the Ariane 6 flight itself, still floating between late 2026 and early 2027; a firm flight number and date will fix it. [1][6] After separation, the telemetry worth following is the spiral, about thirteen months of low-thrust cruise and periodic tracking passes before HENON settles near 0.08 AU and returns its first in-situ field measurements, likely in 2028. [2][3] Beyond the demonstrator, ESA and the consortium describe HENON as a pathfinder for SHIELD, a proposed constellation of four to five such CubeSats spread across the upstream solar wind for continuous, rather than single-point, storm warning. [5][2] If the ion engine, the deep-space link and the 23-gram magnetometer all hold up, the case for that constellation gets a good deal more concrete. [3][5]

01 A transmitter nobody could touch

Voyager 2 left Earth in 1977 with its hardware frozen for good. Its X-band transmitter, the microwave downlink near 8.4 gigahertz, radiated roughly 20 watts, and a 3.66-meter dish focused that power into a narrow beam equivalent to 1.32 megawatts spread in all directions, the strongest deep-space signal then flown. [3] Radio power obeys an unforgiving rule, spreading out as the square of distance, so each planet the probe passed was fainter than the last. Jupiter in 1979, near five times the Earth–Sun distance, allowed 115,200 bits per second. [1] The rate then fell with range, yet it fell far less steeply than the signal did, because the ground kept catching up.

The rates in that table held up only because operators changed everything downstream of the spacecraft. [1][2] At Uranus in 1986 the Deep Space Network first arrayed dishes, combining the 64-meter station at Canberra with two 34-meter antennas and the Parkes 64-meter radio telescope 320 kilometers away, and the probe first compressed its pictures on board, sending mostly the brightness change from pixel to pixel at about 3 bits each instead of a full 8. [5][6] For Neptune in 1989 they went further, enlarging the Goldstone 64-meter dish to 70 meters in 1988 and arraying it with a second Goldstone antenna and the 27 antennas of the Very Large Array in New Mexico, 29 dishes acting as one aperture 151 meters across. [6][3]

02 The hard part was arithmetic in decibels

A link budget is the running account of signal against noise that fixes the highest error-free data rate a channel can carry. At Neptune the power reaching Earth was about 5×10^−21 watts per square meter, so even a 70-meter dish gathered only a wisp of energy against the thermal hiss of the receiver and the sky. [3] Between Jupiter and Neptune the distance grew roughly sixfold, which meant the collected power dropped by about 36 times. [3] Because the spacecraft was fixed, every decibel of recovery had to be found on the ground or wrung from smarter use of each received photon.

Larger and combined apertures supplied the raw gain. The 64-to-70-meter enlargement added about 1.4 decibels per dish, and the full array of a 70-meter, two 34-meter dishes, and the VLA added 5.6 decibels, close to quadrupling the sustainable bit rate. [1][2] Coding supplied the rest. Swapping the older Golay error-correction scheme for a concatenated code, a Reed–Solomon (255,223) outer layer wrapped around a convolutional inner layer, cut the parity overhead from 100 percent to about 20 percent and pushed the error rate from 5×10^−3 down to 10^−6, all while running within roughly 4 decibels of the Shannon limit, the hard mathematical ceiling on error-free rate for a given signal-to-noise. [2][3]

03 Why it mattered, measurably

One ratio captures the achievement. A network left at its 1979 capability would have returned about 3,200 bits per second from Neptune, while the actual arrayed and recoded system returned 21,600, a factor of 6.8 that came entirely from Earth. [1][2] Seen the other way, the signal at Neptune was about 36 times weaker than at Jupiter, yet the data rate fell only about 5.3 times, from 115,200 to 21,600, so the ground work erased most of the distance penalty. [1][3] Two of these methods outlived the flyby. The concatenated Reed–Solomon and convolutional scheme became an international CCSDS standard developed out of the Voyager program and was adopted by later deep-space missions. [4] Antenna arraying, first stretched to intercontinental baselines for these encounters, became routine practice and later helped rescue Galileo after its main antenna failed to unfurl at Jupiter. [4][6] The counterfactual deserves candor. The 70-meter dishes and arraying served many programs and would have been built regardless, but a fixed spacecraft receding on a fixed timetable forced the schedule and proved the techniques at record range and record faintness. [6]

04 Echoes today

The doctrine these encounters forced, improve the receiver rather than the transmitter, is now standard operating practice, and the Deep Space Network still arrays its antennas to hear faint craft while the coding lineage that began with Voyager's Reed–Solomon underlies essentially every deep-space link flown since. [4][6] Voyager 2 itself is now more than a hundred times farther from the Sun than Earth, still heard on the same 20-watt transmitter by ground stations that keep getting better. [3]