Possible thrust measured

A project log for EMDrive/satellite

Developing a small fuelless microwave thruster

Paul KocylaPaul Kocyla 06/10/2015 at 19:3431 Comments

EMdrive test results. Looks plausible - i think we have thrust :)
The platform was initially spun counterclockwise. After a whike (indicated by the red bar) the thruster was switched on.
Left: Thruster oriented retrograde (increasing rotation rate decay)
Right: Prograde (decreasing rotation rate decay)

The data of higher rotation rates is a bit jumpy. The reason is that to get the rotation rate without much noise, i sample the duration between the peaks of the platform´s imbalance during rotation. The shorter they are the less precise the measurement - but the trend is clearly visible.

To get the proper time, devide X by 16.

Changing from retrograde to prograde was obtained just by turning the cavity around. Cables, batteries and other componets remained in the same place and orientation.


lucassiglo21 wrote 06/13/2015 at 01:18 point

i know it may be a little boring, but you should do the test several times in both directions, to see if the tendency in the data repeats. i'd really like to see that, and a control test, with no power, several times, just to see how is the normal braking of this thing. 
if you manage to get similar initial conditions between several measurements and the tests with and without power, you may average them and substract the control to the measurements with power.
or you can do one test without power, and immediately one with power (avoid thermal/pressure differences this way, at least in a gross way), then subtract them. repeat this 5-10 times if it looks promising, or very not promising.
try to get the same initial conditions each time so you can compare, this can be tricky. may be a good idea to semi automate it.
if you get noise, random results, publish it anyway, it is as important as a positive result.

  Are you sure? yes | no

lucassiglo21 wrote 06/13/2015 at 01:21 point

by the way, i am an experimental physicist, i am really interested in this thing, ask for help if you need. 

  Are you sure? yes | no

Paul Kocyla wrote 06/14/2015 at 16:43 point

More tests will come. Spent a whole day just preparing the first one. Had no more batteries for another.
Now i know how the rig behaves, so I can make some repetitions.

  Are you sure? yes | no

Marvin Macportain wrote 06/12/2015 at 13:14 point

Hi, I'm from Aachen too! The subreddit is hungry for much more information than you're giving them.
Would it be OK for me to meet your team and check out the device, for documentation purposes? I'm a scientist myself and could help with setting up experiments and analyzing the data.

  Are you sure? yes | no

Paul Kocyla wrote 06/13/2015 at 16:13 point

Hi. I know, the problem is I have a full time job, a wife and a handfull of projects running simultanously. We give information out that we have and no time to make a perfect data analysis or fancy cut videos.
But to face it: We are doing it as a project because we are fascinated about it and just share what we did, not more. I try to ignore some flamewars about how "amateurish" our approaches are or whatever, thats bullshit from people who never try and never do, crocodile mouths. If you want, we can meet in Aachen at "digitac" ( on tuesday, I would be happy to give u more information and drink a beer together with you. Cheers

  Are you sure? yes | no

Marvin Macportain wrote 06/14/2015 at 13:31 point

Tuesday it is! I'm sure we can find a couple of very low-effort and high-impact improvements.

  Are you sure? yes | no

Paul Kocyla wrote 06/14/2015 at 16:38 point

20:00h ok?

  Are you sure? yes | no

Tony May wrote 06/12/2015 at 04:53 point

Your test is very clear but the graph is at the same time confusing. Retrograde when powered it slows faster and the drag seems to damp out the noise, to me that makes since.  Prograde when powered rotation slows slower, noise continues it rotates more freely. I tend to think mechanicly so these to different events makes since. My confusion comes because it's not clear if the power comes on at the same speed of rotation, I'm just assuming it dose.     

  Are you sure? yes | no

Paul Kocyla wrote 06/14/2015 at 17:16 point

This noise in the graph was because i measured the rotation by the distance of the imbalance peaks. The faster it rotated, the less precise i could measure the distance so it looked more noisy. The new test is much better.

  Are you sure? yes | no

Matt Papageorge wrote 06/11/2015 at 18:47 point
Cool Setup! 

While this is good news if you are right, I am skeptical that these graphs show thrust. I know that the higher frequency test with this model was a long shot to begin with, so if it doesnt work the effect is not necessarily busted. After all we are shooting into the dark until someone comes up with a theory explaining the effect.

So, as I understand it the experiment spins the drive in the
same direction for each graph, but you thrust in an opposite ways for each. If you were to draw a pair of trend lines through both of those curves using the points average slope before and after the red line I do not know that they would be significantly different. You need a null test case, because a decent amount of friction is clearly present in both of these cases.

I am sure you plan to gather this data if you have not
already (I am assuming this is preliminary results) Looking forward to seeing

Also feel free to correct me if my analysis is wrong.

  Are you sure? yes | no

Paul Kocyla wrote 06/14/2015 at 17:18 point

It´s right - more tests are on the way :)

  Are you sure? yes | no

peter jansen wrote 06/11/2015 at 18:32 point

With all due respect, this is extremely poor science.  The first duty of every scientist is not to fool themselves, and especially not others, and reporting thrust based on two shaky graphs which is now being reported on reddit (and probably soon to make the rounds to every blog that doesn't have a background in statistics or experimental design) is definitely not okay.  Even assuming this device is capable of providing thrust -- and I'm of the camp that /wants/ to believe, but clearly needs to see a compelling demonstration of this -- there are very obvious problems here.  There are only two samples, and they're both poorly controlled and extremely noisy!  Where are the multiple runs, the data to calculate the variance, the statistics, the /baseline/? Why does the retrograde appear to show a change in rotation rate immediately, but it takes the prograde 7000 samples (~400 seconds) to begin to show a change in rotation rate?   Why were they started at different velocities, and different times?  Those are just the very basic questions to begin to see if there is /some/ thrust from /somewhere/, and to begin to characterize it, let alone determining where that thrust is coming from (some part of your experimental setup, or the thruster itself). 

  Are you sure? yes | no

Benchoff wrote 06/11/2015 at 20:26 point

Seconding this. This project is going to need some serious, serious stats work. This project really, really needs someone who knows R.

If the project creators want, I'd be more than happy to put a call out for people familiar with stats and experimental design out on the blog.

Actually, after seeing that you're not taking any more contributors for the project, I'm going to say this in an official, 'I'm kinda running this contest' capacity: You need someone familiar with stats and experimental design on your team.

I'm all for replication, and even if the results you ultimately get aren't  significant, you're already doing great work. You're already doing 90% of the work to make this really, really interesting, but right now anyone can dismiss it out of hand with the current methodology and complete lack of any math. Please, please find someone who knows R and how to run this experiment. 

  Are you sure? yes | no

Alex Rich wrote 06/12/2015 at 00:01 point

I would almost say stats is overkill, as is the vacuum chamber.  Drag force is zero a until you are moving, so why not set this up with everything stationary, don't bother with vacuum, turn it on, see if it moves.... Is that a stupid idea?  To me having this thing  spinning in an imperfect vacuum and measuring changes in rotational speed seems like it is needlessly complex.  Another gentle criticism I have is that we should avoid jargon like retrograde and prograde, etc.  It only confuses the facts and forces people to Google every other word of the writeup.  "Thruster pointed in direction of rotation" is better and more accessible than "prograde."

  Are you sure? yes | no

Victor O. Santos Uceta wrote 06/12/2015 at 15:40 point

Good advice  Dr. Jansen, as a PhD student I see clearly your concerns. But I think this is not poor science, actually, it's just an independent project and not exactly academic research. After all one of the reasons people are trying to replicate this by themselves is because Journals and most physicists don't want to get into the thing, informally trying to figure out what is happening and finally causing a huge mess like the "warp drive" commotion. But yes, they should even provide the data and a lot of samples in order to make such claims.

  Are you sure? yes | no

Paul Kocyla wrote 06/14/2015 at 17:41 point

What a nice compliment to call it science - no matter if poor or not :-D
Hey, it´s just a garage project. And after the recent tests it really seems to work, so I don´t really understand the problem. We do it, we share it, we have fun and excitement doing it. We don´t need any supervisor watching down on us telling us how things have to be done. If you can do it better - feel free to do so. Consctructive comments are of course always welcome, and we use the hints and ideas for future tests, and we keep sharing what we have - what some "real" scientists often don´t.

  Are you sure? yes | no

K wrote 06/11/2015 at 09:31 point

Additional question - how important is surface quality of the cavity? On the picture it looks extremely bumpy, maybe it's better to polish it to near mirror-like level?

  Are you sure? yes | no

Paul Kocyla wrote 06/11/2015 at 09:56 point

Thanx for the hint. We´ll give it a try.

  Are you sure? yes | no

concretedog wrote 06/11/2015 at 10:39 point

The surface finish wasn't too bad actually in the cone.. the bumpiness you're seeing is just the machining markings rather than ridges.. but I will attempt a higher quality finish on the next one.

  Are you sure? yes | no

K wrote 06/11/2015 at 08:07 point

Would be nice to create enclosure from which you'd be able to remove most of the air. Next, initiate spin using external electromagnet pulses. Record the curve of the decay (chart #1). Next, Restart the spinning, this time in the middle of recording (chart #2), turn on the EMD via laser beam (you probably would have to install sensor and indicator for that). Comparison of curve #1 and #2 would be great!

Great work!

  Are you sure? yes | no

Paul Kocyla wrote 06/11/2015 at 09:02 point

Thanx. We got a vacuum chamber, but the levitator is too high, so we need to modify it.
The thruster can be turned on by radio.

  Are you sure? yes | no

Tony May wrote 06/11/2015 at 03:58 point


I've been watching the threads on this and I now have reason to think it's not been a wasted effort.

I also wish you the best!

  Are you sure? yes | no

Paul Kocyla wrote 06/11/2015 at 09:03 point

Thank you. We will continue experimenting and posting results.

  Are you sure? yes | no

Mike McCulloch wrote 06/10/2015 at 21:23 point

Neat & elegant experiment. Please can you clarify what your x/y axis units are, your power input & cavity dimensions? Thanks. Best of luck!

  Are you sure? yes | no

Paul Kocyla wrote 06/11/2015 at 09:59 point

X is time in seconds when you divide by 16. Y is the rotation rate (unit not specified, I´ll need to lookup the factor) We´ll post the details of the cavity later

  Are you sure? yes | no

John Romero wrote 06/10/2015 at 20:36 point


A request: can you publish the slowdown curve of a null test (i.e. a spun up but unpowered Emdrive rotating alone until it stops)?

If would be interesting to have such a curve for comparison purposes.

Keep it up.

  Are you sure? yes | no

esot.eric wrote 06/11/2015 at 01:14 point

wouldn't such be affected by things like the phase/position of the moon...?

  Are you sure? yes | no

Paul Kocyla wrote 06/11/2015 at 10:02 point

Probably not. Maybe in long time measurements earth rotation and gravity change due to tidal effect would play into the measurements, but they should be within the noise.

  Are you sure? yes | no

Tony L wrote 06/10/2015 at 19:46 point


  Are you sure? yes | no

Paul Kocyla wrote 06/10/2015 at 19:53 point

We must still verify it by more tests, but it looks promising :)

  Are you sure? yes | no

AlainCo wrote 06/10/2015 at 20:29 point

interesting setup.

To eliminate artifact, you could ask people like shawyer, yang juan, Nasa EW, check in emdrive FAQ...

note that you should also see the importance of pressure, testing for example with half or double (low) pressure, to see if it is source of thrust.

one idea also is to install 2 emdrive and switch them...

wish you the best!

  Are you sure? yes | no