Impossible Thruster Probably Impossible

I’ve gotten a few queries about this “Impossible space drive” thing that has space enthusiasts all a-twitter. This supposedly generates thrust through the interaction of an RF cavity with a “quantum vacuum virtual plasma,” which is certainly a collection of four words that turn up in physics papers. An experiment at a NASA lab has apparently tested a couple of these gadgets, and claimed to see thrust being produced. Which has a lot of people booking tickets on the Mars mission that this supposedly enables.

Most physicists I know have reacted to this with some linear combination of “heavy sigh” and “eye roll.” The proposed mechanism doesn’t really make any sense, and more importantly, even in the free abstract for their conference talk they state that both the configuration of the device that was supposed to produce thrust and the “null” version that was not supposed to produce thrust gave basically the same result. As Tom notes, this is mind-boggling, and John Baez goes into more detail, including a link to the paper.

The paper itself is kind of a strange read, like it was put together by a committee containing a mix of responsible, hard-headed engineers and wild-eyed enthusiasts. The experimental procedure and results sections are very sober and pretty clear that this is not a meaningful test of anything, but then there’s a whole section planning missions to Mars with scaled-up versions of the technology. Which sort of suggests that this was a test run by some career engineers at the insistence of an enthusiast who’s highly-placed enough to make them do tests and write up stuff that they find kind of dubious. But that’s just speculation on my part.

The only thing I have to add to this discussion is a quick mention of why this is likely to have gone wrong. The core technique described in the report is a “torsion pendulum.” This is a technique for measuring tiny forces that dates back to the days of the singularly odd Henry Cavendish, and is still one of the principal techniques for measuring the force of gravity. The basic idea is to hang your test system from a thin wire, balanced at one end of a barbell-like arm, then do something that makes the barbell twist. The amount of twist in the wire will then tell you how much force was produced.

The basic technique has a long and distinguished history. It’s also notoriously finnicky, which is why there’s still a lot of uncertainty and debate about gravity measurements. From stuff quoted by Baez, this seems to be the first use of the NASA lab’s torsion pendulum apparatus, which is not terribly promising. There are zillions of ways this could go wrong, and you’re not going to account for all of them the first time out of the gate.

To give you an idea of what’s involved, one of the very best groups in the world at doing this sort of measurement is the “Eöt-Wash Group” at the University of Washington, whose short-range tests of Newton’s inverse-square law provide the extremely shiny photograph in the “featured image” up at the top of this post. I’ve seen numerous talks by these guys, who are awesome, and in many of them they show a photograph of the lab, which contains a big shiny vacuum chamber and set of magnetic shield at one side of the room, and a knee-high stack of lead bricks right in the middle of the floor. That’s not because some grad student got tired before getting all the lead back to the storage room– the pile is placed very deliberately to counter the gravitational attraction of a large hill behind the physics building there.

That’s the level of perturbation you need to account for when you’re doing these sorts of experiments right. Now, the Eöt-Wash crew are looking for much smaller forces than the rocket scientists in Houston, and Houston is pretty flat, anyway, so they may not need to worry about carefully placing lead bricks. But there are dozens of tiny perturbations that are really hard to sort out– the report specifically mentions vibrations caused by waves in the ocean a few miles away, and if they’re seeing that, they’re going to be bothered by a lot of other stuff. This isn’t something you’re going to sort out in the roughly one week of testing that they actually did.

So, yeah, don’t go booking yourself a ticket to Mars because of this story. It’s almost certainly an experimental error of some sort, most likely a thermal air current due to uneven heating. Which is a failure mode with a long and distinguished history– Cavendish himself noted in 1798 that an experimenter standing near the case could drive air currents that would deflect the pendulum, so he put the entire apparatus in a shed, and took his readings with a telescope. And in his final set of data, he found that he needed to account for the difference in heating and cooling rates between his metal test masses and the wood and glass of the case.

The good news is that there’s enough sober and practical content in the report to suggest that somebody there will eventually do this right. At which time the effect will probably disappear– it’s already a few orders of magnitude smaller than an earlier claim, according to the story linked above. Removing air currents as an issue (which they can do, but didn’t because they were using cheap RF amplifiers that couldn’t handle vacuum) will probably wipe it out completely.

So, don’t go booking tickets to Mars. But do go look at the Eöt-Wash experiment, because they’re awesome, and check out the Physics Today story on measurements of “big G”, because it’s fascinating.

(Also, my forthcoming book has a big section on Cavendish. But that’s not out until December…)

23 thoughts on “Impossible Thruster Probably Impossible

  1. The thing that baffles me most, here, is the breathless acceptance in some quarters _even though the null device generated thrust!!_ Is this not strongly suggestive that the thrust was produced by the outgassing of bullshit or some other similar mechanism?

    And the thing I find most boggling, rather than baffling, is that NASA apparently couldn’t be arsed to find a better grade of capacitor for their RF amplifiers. They’re NASA, for Chrissakes, they do actually know how to make vacuum hardened equipment. I know, I’ve done it for them, and sat through the exceedingly painful design reviews.

  2. My last assignment with the Air Force was with a group charged with attempting to forecast the social & political world landscape 20-25 years out and making technology investment recommendations. I’m not a physicist (my PhD is in computer science), but my BS was in EE so I could at least mouth-out the words. Everyone else’s degrees were in the humanities, so it was up to me to throw cold water on the what-if fairy from time to time. At one point I was asked to look at the EmDrive based on the limited literature available at the time. My verbal report was along the lines of (cutting out the elimination of possible explanations for its apparent performance) “I can find no theoretical basis for this device to work as claimed. If it does work, as the Boeing video appears to show, then facts always trump theory. At the risk of sounding like Lord Kelvin, I’m more inclined to believe that the facts are wrong than that the theory is wrong — without more information we can’t really make that determination. I recommend that we don’t include this in the 2035 toolbox. As I’ve said before, Blue Horizons loses credibility if we include anticipated technologies that violate our current understanding of physics.”

  3. Interesting. A close friend who’s smarter than I am, sent me a link to the abstract, but didn’t comment on it. What struck me was that the two test devices each appeared to produce an effect, and this was interpreted as supporting the hypothesis. I took that on face value, “after all, NASA,” and also took the opportunity to speculate about electric propulsion in space.

    Now it turns out there are good reasons for skepticism, so I’ll re-file this under “reported effect may disappear with further testing.”

    To reason backward from the conclusion, I would say that if both test devices produced identical or very similar results, that by itself should be good cause for skepticism. Reason being, even if both devices produce an effect, the difference in construction should produce a difference in the degree of effect. Embarrassing to admit not seeing that when I read the abstract, but potentially useful as a lesson for avoiding that particular type of mistake again.

  4. Even if the device did work, they talked about getting energy from solar panels,. Did they forget how liitle solar power they’d get on an intertsellar trip as the ship left the vicinity of the sun(inverse square).

  5. I should note that it’s not too hard to retcon the “null” thruster also producing thrust– the design was apparently a modification of one end that’s different from some previous unit that supposedly produced thrust, and was supposed to be a big improvement. If you (want to) believe the result, it’s easy to write that off as “well, the theory that suggested this as an improvement needs more work.”

    I agree with the general consensus, though, that this is vastly more likely to be a signature of some garbage thermal effect than New Physics in action. I can see, though, how somebody who believed in the validity of the underlying mechanism might be able to rationalize reporting thrust from both units as not completely idiotic. (And thus, they won’t be convinced by “the null unit showed the same effect” as a counter-argument…)

    If you look at the report, too, the number of trials is ridiculously low– it’s something like five tests of each unit in one orientation, and only one in the other. Another easy rationalization would be that there will turn out to be some difference between the two, once the statistics get better.


    Said “drive” was never NASA-tested in hard vacuum. When Mythbusters evacuated the ionodrive, it flopped. Make a little aluminum foil swastika with a dimple at its center of mass. Balance it on an upright nail point. Connect to a Tesla coil. Wheee! It spins in vacuum, too, via cold cathode emission. Radiometer, etc. In all cases it is reaction thrust.

    Linear momentum is conserved by the homogeneity of the vacuum plus Noether’s theorems. Immersed in a gravitational potential, vacuum is not homogeneous. Vacuum loopholes are not exploitable as reductions to practice. arXiv:1103.5222, 1107.2886, /gr-qc/0205059 are reaction thrust. arXiv:1402.5022, 1107.5938 close, but no cigar in vacuum.

    Angular momentum non-conservation may have Milgrom acceleration shaping spiral galaxies. 1.2×10^(-10) m/s^2 is a very slow way to travel.

  7. Reminds me of the coherent neutrino detector proposed at U of MD 30 years ago. Similar experimental procedures with similar problems of isolating all outside influences. Congress earmarked funding to support this over and over. Each agency in turn tried to kill it but it kept turning up elsewhere. This in spite of a theoretical calculation by Freeman Dyson showing that if a coherent neutrino detector were possible it would require integration times longer than the age of the universe.

  8. Note that this appears to be the same group at NASA that got some press recently for working on warp drive engines and actually displaying a “mockup” of a proposed starship! Doesn’t NASA see that this kind of thing is embarrassing?

  9. This group at NASA clearly has some issues, as Lee points out. When I saw a headline about this, I knew before reading the article that it was going to be from that group. My real problem with all this is, when pathological work makes a big press splash, it makes the public less able to discern good work from bad. It pains me.

  10. Doesn’t NASA see that this kind of thing is embarrassing?

    More relevant question: Does anybody at NASA who is in a position to do anything about this see that this kind of thing is embarrassing?

    If the research group in question has a powerful enough patron in the bureaucracy, or on Capitol Hill, they can get quite silly indeed before anybody feels they have enough political cover to take action. That goes for any bureaucracy: if the people supplying the money want it to happen, it will happen.

  11. I had no idea Sonny White moved into propulsion after grad school (I knew him tangentially from classes). Working with whistlers and Venus lightning is an enigmatic choice for an Masters-level mechanical engineer to choose, not to mention doing a physics PhD to begin with.

    Doug, despite being on Sonny White’s committee, I suspect you don’t remember his defense (the title page says 2008)? In general would you agree that this is pretty far from Venus lightning and whistler work? I’m not seeing a way to tie this to plasmas (though admittedly after several years in industry, I’ve largely forgotten plasma physics).

  12. agm@11: I do plasma physics for a living, and I have never encountered anybody doing quantum plasma physics, as these guys claim to be doing. That isn’t to say there is legitimate work out there, but most of the applications of plasma physics are in either classical or relativistic settings. The part about using RF to excite this alleged quantum plasma may (or may not–I haven’t read the paper) have something to do with whistlers, and whistlers are frequently generated by lightning, but the relationship is tenuous at best.

    Some years ago there was some theoretical and experimental work on a mini-magnetosphere propulsion system. I knew the theory guy on that project. I haven’t heard anything about it in years, so I suspect that the results didn’t justify further work in that direction. But even if that is the case, at least that project rose to the level of being wrong, which this quantum plasma system seems not to have achieved yet.

  13. G@3: “after all, NASA,”

    NASA is not above doing/announcing bone-headed scientific “discoveries”. Take the “arsenic life” debacle, which was done by a NASA research fellow and announced at a NASA press conference. That was another situation where people who were excited about the implications of the discovery announced their results, and people with experience in the field immediately came forward to say “this violates everything we currently understand about how these sorts of systems work – we strongly suspect that you have some sort of experimental error.” At least for the arsenic life paper, that turned out to be the case, where they saw arsenic in DNA because they didn’t wash off the loose stuff well enough. Time will tell if this result also suffers from inadequate experimental proceedure.

  14. agm, I do remember Sonny. I think that all of his “Mach effect warp drive” and “EM Drive” stuff is very far (in multiple ways) from the (completely reasonable, but perhaps unglamorous) thesis topic of lightning on Venus.

  15. Google points me to an article in Forbes which explicitly claims that the name comes from Scotty’s line, “You cannae change the laws of physics,” and not from the ancient Roman town of Cannae as I had initially assumed. The theory guy on this thing is presumably from Italy, the country in which Cannae is located (but I don’t know if he’s from that part of Italy).

  16. Force produced with a load resistor placed in palce of the test article was 9uN (vs. 49uN with the test article). If the effect is due to heat, then the force produced with the resistor should be larger, as the resistor dissipates all of received RF energy as heat. Further, air circulation for the “slotted” test article will be different then for the “unslotted” test article, so produced force should be different — but both test articles produced the same force.

    As for the sea waves and the shape of local gravity fields. The former can be treated as random disturbance, so after doing several measurements the effect will average out. (There were 4 measurements done for one test article and 5 for the other). The latter is not a factor, because it is constant — and there is a graph in the paper showing the displacement value while switching the RF source on and off — the displacement clearly increases when the RF source is on.

  17. A force that would register as thrust needs to be in a particular direction, which implies some uneven heating. producing air currents that yield a consistent force. That’s way more likely with a complicated shape like their tapered RF cavity than a simple 50-ohm dummy load, which is probably more symmetrical (it didn’t specify the form of the dummy load, that I recall). My impression was that the slots or lack thereof were on an interior surface of the cavity, and thus unlikely to have a big effect on the air circulation. I didn’t look all that carefully at the schematics, though.

  18. This “Impossible Drive” is not the first such claim of unconventional physics leading to propulsive, or force, effects. One that caught my attention years ago was the so-called “Impulse Gravity Generator” of Evgeny Podkletnov in 2001. Podkletnov claimed to have observed 1000 g impulses emanating from YBCO superconductors subjected to 2 million volt discharges. But these impulses only lasted for about 1/10,000ths of a second. Understandably no professional physicists took this claim at face value.

    But a few years later a group at the Austrian Research Center (ARC) subjected a niobium ring to 7.33 g’s acceleration, yielding 100 micro-g’s signal. Out of curiosity I did a back of the envelope calculation of how many g’s acceleration a free electron would experience if subjected to the conditions in Podkletnov’s experiment. It turned out the yield of signal versus electron acceleration between Podkletnov’s experiment and the ARC team’s experiment were not proportional, but differed by a factor which was close to the ratio of electron/proton mass. This was kind of a curious result, showing that if one substituted the proton’s mass for the electron’s mass, that there is a very close linearity between applied acceleration and signal yield, between the two experiments.

    The fly in the ointment is that the protons are bound tightly within the nuclei at the superconductor’s lattice sites, so they are not independent masses. Nonetheless, I was sufficiently curious to run experiments of my own, similar to Podkletnov’s, but on a much smaller scale. Only recently with a little electronic wizardry have I been able to isolate the acoustic ‘pop’ from the high voltage discharge through the superconductor, from the hoped-for acceleration signal. At one time I thought I saw something, but that was before I isolated the acoustic signal, although oddly it didn’t show up when the superconductor rose above the critical temperature.

    My hope is a lab, that routinely works with liquid helium, will allow me to shock pieces of niobium, which has 10 times the cooper pair density with high voltage discharges, and see what shows up. Here is my website describing the experiments:

Comments are closed.