A tour of the worldwide state of EP power — drawn from our founder's own survey of the field, with statuses updated to August 2026. Read it for one thing: how long this takes everyone, everywhere.
Gateway's Power and Propulsion Element (PPE) is a 60 kW-class spacecraft that will manoeuvre the lunar station — co-launching with the HALO habitat ahead of Artemis IV (NASA schedule as of mid-2026: launch no later than December 2027; Artemis IV no earlier than September 2028). PPE was originally conceived with chemical propulsion and redesigned around solar-electric propulsion: for comparable manoeuvres it burns roughly 10× less propellant, and in-orbit xenon refuelling is targeted to extend its life.
Its thruster suite (public status, 2025–26): three 12.5 kW AEPS thrusters — the flight system descending from NASA's HERMeS research thruster — plus four Busek BHT-6000 Hall thrusters.
600 V discharge · 589 mN · Isp 2,800 s · throttleable 3–12.5 kW · ~5,000 starts · magnetic shielding stretches channel life from ~10,000 to ~23,000 hours. Published HERMeS/AEPS-class figures, per the survey's sources.
400–800 V · up to 8 kW · 325–449 mN · Isp 2,210–3,060 s · ~25 kg · throttleable 4–10 kW. BHT-8000-class survey figures; Gateway itself flies the 6 kW BHT-6000.
Behind the plasma: four power processing units drive the whole thruster suite from the station's solar bus. The discharge design NASA demonstrated for this class (DM, 2017) is the yardstick below.
Read those numbers again: the converter spans the thruster's whole envelope, survives its arcs, and reports its health. Half of "a thruster" is this box.
For Earth–Mars cargo you need high thrust, high Isp and a huge throttle range at once. Gridded ion thrusters don't scale well with power; clustering many small Hall thrusters costs footprint, dry mass and efficiency. The University of Michigan's X3 nested Hall thruster answers by nesting three discharge channels in one 250 kg machine: a 2–200 kW design envelope, 200–800 V, the largest throttle range of any Hall system — ~102 kW demonstrated in published testing, with 100 kW-class integrated testing under NASA's NextSTEP programme (XR-100, with Aerojet Rocketdyne developing the PPU).
Every DSU is the same 3–13.3 kW, ~95% ZVS building block from the Gateway section. Modularity buys optimal heat spreading, placement freedom, and redundancy — lose a module, salvage the mission. The PPU scales by modularity, and it is half the machine.
Why it is brutal: the nested channels share magnetic circuits, mix materials with different thermal expansion, and the inner rings have nowhere to dump heat — every solved problem uncovers the next one. Hover or tap a ring.
Silicon MOSFETs are near their theoretical limit. GaN FETs switch faster with lower parasitics, tolerate radiation inherently (no gate-oxide layer), block higher voltages, and shrink the magnetics — because electromagnetic components are over 60% of a converter's mass, and their size falls with switching frequency. But GaN has no body diode, so the industry's beloved phase-shifted full bridge stops being the answer:
Phase-shifted full bridge with zero-voltage switching — the workhorse of flight PPUs. Well understood, analytically tractable, proven at power. Leans on the MOSFET body diode that GaN does not have.
Full-bridge LLC resonant conversion: ZVS and ZCS at the resonance point, sinusoidal currents, low EMI, leakage inductance absorbed into a planar transformer. The price: no easy analytical solution — voltage gain is a complex function of frequency and load — and every design is a custom transformer plus iterations.
A 1.2 kW GaN anode supply switching at 1 MHz, ~94% chain efficiency, slot-card modules on a backplane master controller, ~2–2.3 kg projected for a full unit. De-risk objectives — not a product, not flight heritage. That distinction is the whole story of this page.
| GaN-LLC | Si MOSFET-ZVS | |
|---|---|---|
| Specific power | Wins — smaller devices for the same on-resistance, magnetics shrink with frequency | Heavier — magnetics sized for low-frequency switching dominate the mass |
| Complexity | High — 5 V gate margins, layout-critical loops, LLC optimisation, custom transformers, multiple iterations | Moderate — mature design practice, 12 V gates, analytical models |
| Research still required | Substantial — evaluation boards to prototypes, high-frequency isolation schemes, rectifier compatibility | Little — the known path, at a known weight penalty |
The field's direction is set — wide-bandgap, resonant, modular — but nobody gets it free. The know-how is the product.
Every publicly reported Hall-thruster PPU the survey could find, worldwide. Note the column that matters: status. Across two decades of public reporting, flight-model entries are the exception — most of the world's PPU work has stopped at engineering models. This is historical public heritage compiled through ~2022, not a current vendor ranking.
| Unit · maker | Power | Output voltage | Efficiency | Specific mass | Status reported |
|---|---|---|---|---|---|
| PPU NG · Airbus | 5 kW | 300–400 V | 96% | — | EM, 2015 |
| PPU Electro · Airbus | 5 kW | 300–400 V | 95% | — | EM, 2016 |
| PPU Mk1 · Thales Group | 1.5 kW | 350 V | 92.5% | 7.25 kg | FM, 2003 |
| HPPU · Thales Group | 2.5 kW | 350 / 400 V | 95.1% | — | EM, 2004 |
| PPU Mk2 · Thales Group | 2.5 kW | 350 V | 94.4% | — | QM, 2012 |
| PPU · MELCO | 4.5 kW | 350 V | 94.2% | 5.6 kg | EM, 2007 |
| 13 kW PPU · NASA | 13 kW | 300–600 V | 95% | 3.75 kg | DM, 2017 |
| R-400 PPU Mk1 · Rafael | 600 W | 300 V | 91% | 20 kg | FM, 2017 |
| R-400 PPU Mk2 · Rafael | 900 W | 200–300 V | 92% | 11.1 kg | EM, 2019 |
vidyut is not in this table, and should not be: it is a map of legacy agency and prime heritage. Our role sits after it — merchant PPCUs for the next wave of thruster makers, who will not have twenty years to build this table's lessons themselves.
Programme dating per our founder's first-hand experience inside the programme and public reporting; the flight heritage below belongs to ISRO, not to vidyut hardware.
The moral is not about any one agency. It is that even sustained national programmes take decades to mature flight EP — the difficulty is the moat, and trained people are its scarcest output.
Everything above says the same thing three ways: the power unit is half the machine, the qualified supply of them is a short table, and the know-how concentrates in the few teams who have carried one through a flight programme. A new thruster company can spend its first five years rebuilding this page — or buy the converter from people who already lived it, and stay a plasma company. Our own engineering-model direction — lightweight GaN with resonant soft switching — is drawn straight from this landscape.