Once again the impossible drive is in the forefront of public news.

But thus far, I am still unimpressed. NASA seems to have thrown a lot of resources, to test the viability of the EmDrive. But I am not sure we are any closer to knowing if it "works".

Why don't we build a cube sat, launch it into orbit and try and push it out to Pluto?

Seems like we would get much more useful data much quicker.

  • 4
    $\begingroup$ Is this a troll question? "but thus far I'm still unimpressed" - also PHYSICISTS NEVER DESIGN STUFF "cube, push to pluto" style. Everything is the shape/colour/whatever it is for a reason. $\endgroup$
    – Alec Teal
    Nov 9, 2015 at 18:39
  • 1
    $\begingroup$ @AlecTeal I am unimpressed as there are still so many possible explanation for the force that aren't reactionless, for instance, thermal effects on the air. My point is, as long as the emdrive is on earth I will treat it like super luminal neutrinos. If it gets to Pluto, it would be "new physics". Also, could someone remind me how we know lightning is electrical? $\endgroup$
    – Aron
    Nov 9, 2015 at 23:09
  • $\begingroup$ chat started: chat.stackexchange.com/transcript/message/64722356#64722356 $\endgroup$
    – uhoh
    Nov 15, 2023 at 23:20

7 Answers 7


EmDrive takes ~300W. You won't get it from a cubesat. You need over a meter of solar panels, or good 56kg of RTG battery.

It's been tested on Earth, made with materials and electronics meant to work in Earth ambient conditions: temperature, pressure, radiation. Putting things in space is not as simple as loading them onto a rocket. If they are to work, they need

  • a complex thermal management system (dissipating 300W using only radiators will again require a good few m^2 of radiator area),
  • power source and management (no nice 230V out of wall socket) and a thermal management system for that (frozen batteries don't work)
  • all the electronics must be toughened against space radiation (meaning replaced with special chips made in a technology of thick traces, high currents, and a lot of redundancy, so that errant particles don't change the bits). These are complex, expensive and require special expertise, way different than common electronics.
  • There's the matter of pressure - create a neat component safely embedded in epoxy, get a bubble of air trapped, your neat component explodes in orbit.
  • There's the whole telemetry and radio matter - so you sent your neat drive away and nobody knows what happened to it. Space is big.
  • Attitude control and assuring that the drive's thrust is in line with the center of mass. Otherwise instead of flying to Pluto you'll have the fastest-spinning satellite in the world, spewing pieces around as it breaks apart.

...and funds. How many thousands dollars per kilogram? And this thing won't be a cubesat. It will be something of order of tons.

And for what? If it doesn't work, we won't know if it doesn't due to some failure or because it's a faulty concept. Let's first make sure we have something we can send into space.

  • 1
    $\begingroup$ But maybe in a vacuum container aboard the microgravity of the space station? That seems to take care of all the concerns you list. And add human management to simplify and enhance the experiment. Cygnus or Dragon payload doesn't cost more than a few $10,000 per kg,does it. $\endgroup$
    – LocalFluff
    Nov 9, 2015 at 12:20
  • 3
    $\begingroup$ @LocalFluff: The ISS has many "vacuum experiments" compartments. Currently the device is too big though (just look at the photo by the article; it wouldn't even fit through the hatch of a Progress capsule :) Plus it would hardly withstand launch accelerations and vacuum currently. But yes, this is a worthy avenue of research - building a thruster that could be installed on ISS is better than a dedicated probe; you just about halved the headaches. $\endgroup$
    – SF.
    Nov 9, 2015 at 12:58
  • 6
    $\begingroup$ @MichaelKjörling: 1) Telemetry, communication and attitude control 2) on-site fixes, 3) power (and to a degree, cooling), 4) electronics/backend can work inside ISS, only the thruster part needs to be vacuum-proofed, 5) reduced, more manual control systems (doesn't need to be nearly as autonomous), 6) it can be tacked onto a standard delivery mission instead of having a separate launch; 7) with all these cuts the required mass is vastly reduced so cost to orbit is as well. $\endgroup$
    – SF.
    Nov 9, 2015 at 15:23
  • 2
    $\begingroup$ Eagleworks test article (Cannae, the 300W one you refer to) fit inside a 30 inch by 36 inch vacuum chamber and is 11 inches in diameter and 4-5 inches between the ends of the beam pipes. Also, it's entirely possible to get 300W with deployable thin film PVs on a nanosatellite (state of the art is currently quoted at specific energy density of over 1,000 W/kg), tho it wouldn't be a single 10x10x10 cm and 1 kg CubeSat unit. But it probably could be possible stuffing all that into a 3U Cubesat, maybe something similar to Lightsail-1. $\endgroup$
    – TildalWave
    Nov 9, 2015 at 16:37
  • 2
    $\begingroup$ Other than being in freefall testing aboard the ISS does nothing that testing on the ground doesn't do--the real issue is the coupling between the test object and the experimental chamber. $\endgroup$ Nov 10, 2015 at 0:56

We might, it depends on the scale of the project, and if someone proposed a proof of concept mission and is ready to finance it.

For a small scale technology demonstration mission such as the ones regularly performed aboard the ISS, it could be, for example, proposed through CASIS as a Physical and Materials Science R&D project, but it likely won't win any government sponsored grants (competition is fierce).

As for NASA, it depends to what TRL (Technology Readiness Level) can you qualify it. For a full-scale system component, it is currently only at TRL2 and lacks breadboard tests for independent verification and validation (IV&V) of the technology (TRL3), full-scale experiments (TRL4), and validation in representative environment (TRL5), before it can move to prototype demonstration in relevant environment (TRL6) and beyond.

NASA Eagleworks, in the most recent related paper that NTRS will return, in section V. Application of Technology to Space Exploration Missions, interestingly avoids description of small-scale technology demonstration missions and jumps right to potential full-scale application of the technology with a couple of interplanetary mission examples (to Mars and Titan/Enceladus at Saturn). Those serve as demonstration that what currently stands on paper (and has yet to be independently verified) has potential real-world applications, but that's it. Paper's summary clearly states that:

The near term objective is to complete a Q-thruster breadboard test article that is capable of being shipped to other locations which possess the ability to measure low thrust for independent verification and validation (IV&V) of the technology. The current plan is to support an IV&V test campaign at the Glenn Research Center (GRC) using their low thrust torsion pendulum followed by a repeat campaign at the Jet Propulsion Laboratory (JPL) using their low thrust torsion pendulum. The Johns Hopkins University Applied Physics Laboratory has also expressed an interest in performing a Cavendish Balance style test with the IV&V shipset.


The real information, not the "public news", is what NASA uses. Spinning a mundane negative result into hype, or describing the taking of measurements near the noise level of the instruments and then claiming it's a profound and confusing result, does not fool or impress the real engineers.

It does not work.

There is no reason to suppose anything fishy is going on.

Popular-press articles that claim otherwise, and articles written by the cranks themselves, don't change the actual facts even if they create a public mythology.

Now would a similar measurement in space help? A characteristic article I recall described small force measurements seemingly at random including those in the wrong direction or when the machine was off. Tiny effects in the environment and noise in the instruments swamp any real reading. If you tried it in a free orbiting platform, you would also get random changes due to variable atmospheric drag, solar particles, magnetic fields, outgassing of parts, differential cooling, and light pressure, not to mention peturbations from other satellites and bodies in the solar system and irregularities in the Earth (for a good description read up on Gravity Probe B and the drag-free orbit effect). Taking data, whether by careful positioning information or on-board accelerometers, you would have random changes that cannot be controlled for, exactly like the previous bench tests.

And you will still have reports that acceleration during the "control" phase (turned off) or in the wrong direction is somehow mysterious and suggestive, when it actually means that the trial run cannot be distinguished from the control and supports the "does nothing" hypothesis to enough sigmas to rule out any interesting effects.

  • $\begingroup$ A new paper has been published by NASA that conforms the measurements of thrust and also rules out a lot of sources that could affect the measurement: arc.aiaa.org/doi/10.2514/1.B3612. I'm not saying that the EM drive really works, but with this new paper the possibility should totally be considered. $\endgroup$ Nov 27, 2016 at 20:46
  • $\begingroup$ «The requested article is not currently available on this site.» $\endgroup$
    – JDługosz
    Nov 27, 2016 at 23:59
  • 1
    $\begingroup$ JDługosz, cc@JannikPitt: There was a typo; the correct URL is arc.aiaa.org/doi/10.2514/1.B36120. $\endgroup$ Jan 22, 2017 at 0:22

I was just researching this now.

A working design basically doesn't exist yet, so it's still a question of "send what up?" All the major initial results are far more likely the result of thermal currents, which is obviously the case. Since then, the measurements have mostly been within the range of error.

There still appears to be promise, but they're still working on test sensitivity and drive design until they get some results that aren't probably the result of electronic noise.

Also apparently a bunch of electronics that work on earth don't work on space. Who knew?!

  • 1
    $\begingroup$ And some things that work in a microgravity vacuum environment don't work here on the surface on Earth in the presence of an appreciable atmosphere. But when we understand why something works in one environment and not the other, we have far better odds of building something that can actually do what we ask of it. $\endgroup$
    – user
    Sep 28, 2016 at 15:26

Why don't we build a cube sat, launch it into orbit and try and push it out to Pluto? Seems like we would get much more useful data much quicker

Well, ignoring the details of low earth orbit, lack of space for power and problems of monitoring a small dark object over vast distance ...

To traverse 7.5 billion kilometers starting at 0 m/s and using an accelleration of 50 micronewtons on a 1Kg cubesat means you would get your data in about 38 years. That doesn't seem very fast.

  • 5
    $\begingroup$ If it moved at all, you'd have the data you need. It doesn't really have to get all the way there. $\endgroup$
    – Shane
    Nov 9, 2015 at 21:50
  • $\begingroup$ Interesting. This article did not do the math nor give any justification for the 18-month figure. $\endgroup$
    – JDługosz
    Nov 10, 2015 at 14:46

NASA doesn't work like that. It might be quicker, but it's also a lot riskier, and it has very little value. A Cubesat can't carry much instrumentation, so after flying it we'd be no closer to understanding how and why the EM drive works, which is the more important question at the moment.


And in 2018 we found out that the EmDrive doesn't work, without ever having to take it off Earth.
Tests on Earth have some big advantages over tests in space. On Earth, we can easily examine and modify the test system, to progressively rule out errors. We can progressively add more test equipment and e.g. shielding to refine the test.
If you launch an EmDrive on a small satellite, you get only one chance to build a test system. Any follow-ups with a changed test system would be very expensive as they would require another launch.

  • $\begingroup$ Exactly. We would be the embarrassing joke among space faring civilizations if we were to take a step towards space with a device that we are not exactly sure how and why it works. $\endgroup$
    – Mike Nakis
    Nov 9, 2015 at 13:59
  • 2
    $\begingroup$ Don't know how or why but DOES would make the joke worth it. $\endgroup$
    – Joshua
    Nov 9, 2015 at 16:42
  • 3
    $\begingroup$ Humans used steam engines for over a century without understanding exactly how any why steam engines worked. Humans used electrical batteries and vacuum tubes for decades before the 1897 discovery of the electron. Learning exactly how and why something works is certainly nice; achieving useful results is also nice; perhaps it is best to do both in parallel without forcing either one to wait on the other. ( Mike: is that sincere or sarcasm? Sometimes it's hard to tell on the internet. ). $\endgroup$
    – David Cary
    Nov 10, 2015 at 15:54
  • 2
    $\begingroup$ Steam engines were in practical use for many years before they were first used in a transatlantic voyage. They were also well-understood on a practical level. Space is difficult and risky enough without harebrained schemes. You do Earth-bound tests first, THEN you use your new technology in space. Not the other way round. $\endgroup$
    – Hobbes
    Nov 10, 2015 at 18:14
  • $\begingroup$ The electron was discovered 1897 and the Edison-Richardson-effect of vacuum tubes was found between 1873 and 1889. $\endgroup$
    – Uwe
    May 21, 2018 at 11:05

Paul March (if I remember well) explained that just to have Nasa GRC to accept working on EmDrive test, they needed to have reach a minimum thrust (50 or 100µN as I remember).

Getting straight to space with an untested device may lead to a fatal false negative. It happened that way with Cold Fusion, with two influential labs who failed , for reason that were unknown at that time, and because they made bad assumptions and refused to ask to experts (Nasa EW did similar error relative to Shawyer).

For me that false negative risk is a much more serious risk, thank the delay and the budget. If nasa can accept failure without dumping the idea , it is OK, but my perception is that it does not work that way.

  • $\begingroup$ See this book regarding cold fusion in particular. $\endgroup$
    – JDługosz
    Nov 10, 2015 at 13:01
  • $\begingroup$ This is a good book to understand how people can ignore evidence, be so confident in their biased reasoning that they write a book. Gary taubes book, and Huizenga books are also good to learn the popular fallacies among some scientist too focused on theory. The best book to understand how Colf fusion was ignored, learning calorimetry and good epistemilogy is "Excess Heat" by Charles Beaudette. Moreover it is published by Uni Tsinghua as PDF. iccf9.global.tsinghua.edu.cn/lenr%20home%20page/acrobat/… The end is very good on the bad books. $\endgroup$
    – user26277
    Nov 11, 2015 at 19:28

Not the answer you're looking for? Browse other questions tagged or ask your own question.