🚀 Space
Three Satellites Flew in Formation on Magnets Alone. Real Orbit Demands 60 Million Times More Force.
A University of Kentucky team steered three satellites with frequency-multiplexed magnetic fields, no propellant, holding formation to ±1 cm in 22 seconds. Then you run the scaling law: at a real 150-meter baseline their copper coils produce 1.4 piconewtons against an 82-micronewton gravity gradient. The algorithm works. The physics has a price.
Sixty million to one is the ratio between the force needed to hold two satellites in formation at a real mission baseline and the force the newest propellant-free control demo can deliver at that distance. A University of Kentucky team just published the first three-satellite demonstration of electromagnetic formation flying using alternating magnetic field forces: three units on air tracks, pushed and pulled by magnetism alone, settling into formation in under 30 seconds with centimeter precision, no thruster fired once, a closed-loop three-body result nobody had achieved. Control theory, meet hardware: magnetic force between dipoles falls as the fourth power of distance, and 1/r⁴ is a cliff, not a slope. Physics always collects.
Lexington's setup, from Sumit Kamat, Ajin Sunny, T Michael Seigler, and Jesse Hoagg, put a 500-turn copper coil, LiPo batteries, a Copley amplifier, an Arduino Due, and a laser rangefinder on each of three low-friction air-track gliders. Frequency multiplexing does the work: instead of steady DC current, each unit broadcasts a sum of sinusoids, and two units feel a time-averaged force only when their magnetic moments share a frequency. Running 10 Hz and 20 Hz simultaneously, the center unit talked to one neighbor on each channel while the third pair, sharing no frequency, ignored each other, which is the entire problem with three or more satellites, since with two the physics is a parlor trick but with three every field couples to every other field and earlier approaches needed one central controller solving the whole formation at once. Decentralized instead, with each satellite reading only its neighbors, the formation acquired in 22.4 to 25 seconds with maximum error ±0.01 m and mean error under a millimeter. Its lineage runs from a 2021 Kentucky dissertation on frequency multiplexing back to the RINGS experiment that flew superconducting coils on the ISS in 2013, and the funding, NASA, NSF, and the Air Force Office of Scientific Research, with publication in Aerospace Science and Technology, reflects the dual pull of future observatories and resilient military constellations.
Propellant Is the Life Limiter
Propellant runs out, and that sentence has ended more satellite missions than any hardware failure. ESA's Proba-3, the current state of the art in precision formation flying, keeps two spacecraft 150 meters apart with millimetre accuracy at the top of a 60,000-kilometer orbit, precisely because Earth's gravity, atmospheric drag, and magnetic field make formation-keeping propellant-hungry anywhere lower, so every 19.6-hour orbit the formation is broken and re-acquired rather than held through the 600-kilometer perigee. Sweden's PRISMA demo flew autonomous LEO formations from 2010 to 2015 on 11 kilograms of usable propellant, about 110 m/s of lifetime delta-v on its main spacecraft, and one precision orbit-keeping experiment alone burned 0.1347 m/s over 29 days just for absolute control. Scale that to dozens of spacecraft holding station for a decade, which is what distributed-aperture telescopes, gravity-wave detectors like LISA, and surveillance swarms all require, and the propellant budget becomes the mission design. Electromagnetic formation flying attacks exactly this, because the coils run on solar power, which does not deplete, and because thruster plumes no longer foul the sensitive optics riding along.
Running the Math the Paper Skipped
Starting from the paper's own hardware specs, each Kentucky coil carries 500 turns of 22-gauge copper at 16 ohms with a 2.35-amp current limit. Cross-checking resistance against wire gauge gives 301 meters of wire, a coil radius near 0.096 meters, and a magnetic moment around 34 ampere-square-meters per unit. Two coaxial dipoles attract with F = (3μ₀/2π)·μ₁μ₂/r⁴, which at the lab's 0.45-meter baseline is about 17 millinewtons, consistent with the paper's own note that current saturates at 2.35 A during acquisition, since millinewton forces are what it takes to shove the gliders into formation in 22 seconds.
| Baseline | Coil force | What the formation actually fights |
|---|---|---|
| 0.45 m (lab) | ~17 mN | Air-track friction, grams of imbalance |
| 3 m | ~9 µN | Differential LEO drag (~10 µN): parity |
| 10 m | ~69 nN | — |
| 150 m (Proba-3) | ~1.4 pN | Gravity gradient at 60,000 km: ~82 µN |
Stretch the baseline to Proba-3's 150 meters and the same copper coils generate 1.4 piconewtons, while the gravity-gradient differential they would fight, even at 60,000 kilometers where ESA flies to keep disturbances small, is 2GM·d/r³, roughly 4×10⁻⁷ m/s² across 150 meters, or 82 micronewtons on the 200-kilogram occulter: sixty million to one. Matching 82 micronewtons at 150 meters would demand a magnetic moment of 2.6×10⁵ A·m² per satellite, about 7,700 times the Kentucky hardware, which no copper winding achieves; that is superconducting-coil territory, with cryogenics, quench protection, mass, and power budgets the air-track demo never had to carry. Their control algorithm would transfer to superconducting hardware unchanged, but the magnet is not a detail, it is the mission. It never was.
Where the Physics Actually Works
A 3-meter baseline is the version of this story where copper works: at that separation the Kentucky coils make about 9 micronewtons, the same order as the differential atmospheric drag, near 10 µN, between two 1-square-meter satellites at 500 km altitude. Close LEO pairs fighting drag are exactly where propellant limits mission life and where the plume-contamination argument bites hardest for optical payloads. One more catch, flagged in the paper as future work, is serious: Earth's magnetic field at LEO runs about 50 microtesla, which torques a 34 A·m² coil at roughly 1.7 millinewton-meters, enormous next to the micronewton-meter disturbances small satellites normally see. Every magnetic push between two satellites also twists both against the planet, so reaction wheels or magnetic torquers must eat that torque, and the companion paper the same group submitted to IEEE, adding control barrier functions for state and input constraints, is really about managing limits like these.
The Strongest Case Against This Article
Stated at full strength: this piece holds a control-theory result to a hardware standard it never claimed, since the authors' contribution is the first decentralized 3+ satellite AMFF demonstration, an algorithm that runs on any coil technology, copper today and superconducting tomorrow, making the piconewton arithmetic a design input rather than a refutation. Nobody proposed flying the air-track units, EMFF was never pitched at Proba-3's 150-meter geometry, and its natural home is close formations where 1/r⁴ is kinder, propellant savings are real, and every caveat, one dimension, lab hardware, thrusters still needed for orbit maintenance, was stated up front. Judging a mathematics result by force output is like judging a compiler by the laptop it shipped on. Numbers don't flatter. Both claims survive contact: the contribution is real and the wall is real, and a field that cannot quote its own scaling law will overpromise, because a three-satellite demo settling in 22 seconds looks flight-ready in a way the equations say it is not.
What This Does Not Prove
My dipole extrapolation uses the coaxial-alignment maximum and the paper's own dipole model, so treat the 150-meter figure as an order-of-magnitude illustration, not a mission design; coil geometry and waveform details move it by factors of a few, not by the seven orders of magnitude separating the demo from the mission. Their experiment is one-dimensional: no attitude coupling, no Earth field, no radiation, no thermal swings, no spaceflight, and while The Debrief reports the September 2026 journal issue, I worked from the arXiv preprint (v4, May 2026) listed as the submitted manuscript. For the Proba-3 comparison, the gravity-gradient figure assumes the gradient dominates at apogee; solar pressure and residual drag add smaller terms that leave the ratio essentially unchanged. Finally, 1.4 piconewtons is the single-pair maximum; a real controller shares force across neighbors, so flight hardware would deliver less, not more. Physics does not negotiate.
What You Can Do
If you evaluate formation-flying proposals, judge EMFF pitches on two axes, coil technology and baseline distance: copper coils past ~10 meters fight 1/r⁴ and lose, while superconducting coils or baselines under ~5 meters are where the physics permits the economics. Ask for the magnetic moment in A·m² and the worst-case differential disturbance in the target orbit, then do the division yourself; it takes one line and it separates hardware from hype. Satellite builders should look at drag makeup for close LEO pairs, where 9 µN-class forces match the disturbance budget and the no-plume argument is strongest. And if you track distributed space telescopes, watch the RINGS lineage rather than the air tracks: the milestone that matters next is a superconducting three-satellite demo, because that is the hardware the algorithm has been waiting for.
The Bottom Line
A Kentucky lab solved the hardest part of the control problem, steering three satellites with magnetism alone into formation in 22 seconds to within a centimeter, decentralized and propellant-free, a genuine first worth everything the authors claim. Control is the cheap half of electromagnetic formation flying; the expensive half is the magnet, because force falls as the fourth power of distance and real missions fly 150 meters apart, not half a meter, which leaves the demo's copper coils sixty million times too weak and points the fix at superconducting coils with cryogenics and quench protection, technology RINGS flew thirteen years ago that nobody has productized since. The algorithm is solved, the magnet is the product, and capital should flow accordingly. Fund the magnet.