I can’t help but be excited about the EmDrive, the experimental space propulsion system that seems to defy the fundamental laws of physics. A peer-reviewed paper has just been released, and, despite many physicists expecting this paper to finally kill the EmDrive puzzle, the opposite has happened: the paper found that the drive does, in fact produce thrust. It’s just that nobody seems to know how. Or why. I asked our tame physicist to help us figure this out.


For those few of you not building interplanetary spacecraft in your backyard from a pair of old Econolines and an Opel GT, some review of the EmDrive is in order.

The EmDrive is a type of RF Resonant Cavity Thruster, which is a fancy name for what is essentially a can shaped like a truncated cone, which uses a source of electrical power to bounce microwaves around inside the chamber.

Somehow, this bouncing of microwaves around inside the chamber is producing a measurable thrust–at least that’s what this recent, peer-reviewed paper found.


This should be impossible, because it violates the idea of action and reaction that has been the basis for every space propulsion system to date; the way the EmDrive seems to work, it’d be like moving a van by bouncing basketballs off the dashboard from the inside. It just doesn’t jibe with physics as we understand it.

The EmDrive does NOT work like this.

Still, a real, peer-reviewed paper has measured thrust from the engine. Not a lot of thrust, mind you, even significantly less than low-thrust propulsion systems like an ion drive engine, but thrust, and consistent thrust nevertheless. All without propellants or fuel, just electrical power.

I have no idea what’s going on here, because I’m what neurologists term “sorta stupid.” Happily, I know people who don’t suffer from this, like our captive physicist, Dr. Stephen Granade.



I asked Dr. Granade to explain to me what he thought was happening, and if this baffling machine actually works. Here’s what he said:

The new NASA Eagleworks paper on the EmDrive is out, and it’s causing one heck of a stir. I’m intrigued by their results, but I’m still far from convinced that the EmDrive is real. The next thing I’d like to see is a different independent lab try to reproduce the results with a different experimental setup.


First, the good stuff. The new test addressed a bunch of my concerns about their test from two years ago. They performed the test in vacuum, and they had a “null test article” that wasn’t supposed to show any thrust and that didn’t, unlike last time. Those are important steps in knocking down possible errors from how the experiment was set up.

However, there are still several issues I’d like to see better addressed. The big one is how you tell the difference between any actual thrust from the drive and the effect of the drive heating up because you’re dumping a lot of electrical power into it, because their measurement technique doesn’t clearly separate the two.

The amount of thrust that EmDrive is supposed to produce is tiny. To measure it accurately, the Eagleworks team used a torsion balance, which is a fancy way of measuring really weak forces. A torsion balance is a bar dangling from a strong fiber. If you push on the bar, it spins, twisting the fiber. But the more you twist the fiber, the more it resists being twisted, like how a spring pushes back harder the more you compress it.


Eventually you reach the point where the fiber’s resistance to twisting is as strong as the force pushing on the bar, and the bar stops turning. By measuring how far the bar rotated, you can know how much force is pressing on the bar. So in theory, all you have to do is strap the EmDrive to the bar, turn it on, and measure how far the bar turns to know how much thrust, if any, the drive is producing.

Scientists have used this setup since the late 1700s to measure all kinds of small forces, like how strong the attraction is between charged particles and how hard light pushes on things. For the torsion balance to work, though, the thing strapped to the bar can’t move. But the EmDrive gets very hot when it’s operating. In NASA’s setup, the EmDrive dumped a lot of that heat into an aluminum heat sink on the back of the device. As the heat sink got hotter, it expanded, and changed the distribution of mass. That moved its center of gravity...and made the torsion balance turn. It’s like sitting on a playground swing and then leaning back. The swing will push forward.


To deal with this, the Eagleworks team tried to separate the two effects in the signals they recorded. They made some calibration measurements using the attraction between electrically-charged particles and, based on that calibration, made assumptions of what the signal due to the thermal expansion would look like. But the torsion balance’s response time to the calibration signal was very different than what they saw when they turned on the device. That makes me do a bit of a Fry squint at their results.

I’d love to see a different version of their null test. Their null test involved pointing the EmDrive along the beam instead of at right angles to it, so that any thrust it might produce doesn’t twist the beam. However, that may have changed what the device’s thermal expansion did to the signal. I’d really like to see a test where they use a version of the drive that’s just a plain old cylinder like a soup can, not tapered like the EmDrive is. The claim is that the taper is what makes the thrust occur. In theory, a version of it without the taper wouldn’t produce any thrust.

Speaking of the force they measured, their measurements varied a bunch from run to run. When they put 40 W of power in, they got a force of 40 µN…plus or minus 20 µN. That’s an uncertainty of 50%. The 60 W power result had an even larger relative uncertainty.


For me, all of this points out the need for a set of independent tests from a different lab with a different setup. These results may show that the device is producing thrust…but it may not. The Eagleworks results are interesting, but they don’t prove that the EmDrive works.

We also need other tests because previous EmDrive tests have been inconclusive, unfortunately. For instance, a Chinese team reported observing thrust from the EmDrive back in 2012, but when they refined their experimental technique, the thrust vanished.

If the EmDrive does work, and it actually doesn’t shove either light or matter out the back of the rocket, then I’m not that interested in it as a space drive. The NASA Eagleworks results showed that the drive produced less than 10% of the thrust per kilowatt of electrical power that an ion drive like NASA’s NSTAR or NEXT drive does. And even if the EmDrive isn’t using propellant, it still needs a power plant to produce the electricity, and that will use some kind of fuel. Even if you want to use solar panels so that you’re not carrying your power source with you, if you want to fly out of the solar system with the EmDrive, your sunlight is going to drop off very quickly.

What I would use the EmDrive for is as a limitless energy generator. The EmDrive is supposed to generate a constant force pushing a spaceship forward. As the spaceship speeds up, it accumulates kinetic energy, the energy of objects in motion. As the spaceship goes faster and faster, its kinetic energy increases as the square of its velocity.



If the EmDrive stays exactly as efficient at turning electrical energy into thrust no matter how fast it’s going, which is the claim, then at some point you’ll have more kinetic energy in the system than electrical energy that you’ve put into the system.

Once you’ve got a system that produces more energy than you put into it, you have a perpetual motion machine. And once you have that, you have a way of generating as much energy as you like.

So, there you go. Our physicist is skeptical, and thinks that even if it does work, it’s not a practical spacecraft drive. But, it would make a fantastic other impossible machine, a perpetual motion machine.


Nothing makes any sense anymore.