The Flight Thruster had reported thrust of 170mN at 440W input and a cavity Q of 50,000. As the Shawyer thrust equation should scale with power, Thrust = (2*Q*Df*P)/c, the Baby EmDrive at the reported 40mW (0.04W) of Rf power should produce 15.4uN of thrust. That is about 50% the weight of a single SnowFlake being 0.003g.

How do you expect to see / measure thrust that is the equivalent of 1/2 a SnowFlake falling on your test rig?

I have been reviewing these experiments for some time. In torsion test #2 and #3 there was probably a signal corresponding to the EMdrive being activated. The problem is that the thrust signal is weak compared to the noise. Given that, I think the approach being adopted right now to test the device with different set ups makes a lot of sense. You guys should continue to vary the set up to see if you can achieve better signal to noise ratio. The idea of using oil instead of water has been suggested and that is just a small change which might yield some results. Also, the test set up with the magnetically levitated test rig was quite brilliant and so far has yielded the most interesting results. It might be worthwhile to make further efforts to ensure the weight is balanced and perhaps test the device in partial vacuum on the levitated platform. The idea of assembling a platform floating on water also sounds promising.

The truth is, that given the small amount of thrust produced, the only way to identify the thrust likely will be to do a large amount of runs and assess the results statistically.

It's not surprising that the experiment is not showing much thrust. A number of experiments already completed with the EMdrive show that thrust increases with power. The experiments completed by Juan and Wang in China actually show that reported thrust increases in a non-linear fashion. So you have you work cut out using such low power, but some of the results you have presented so far are already suggesting thrust, even if the signal is very weak.

I also think it great the that your team has released information on these experiments on a regular basis. Other experimenter can learn what works and what does not. Even if a test shows nothing, at least we have learned that that specific set up is not ideal to test the EMdrive. I am sure by the end of the process we will know what testing method works best.

Thanx, that´s a perfect summary.
After building something new, we wanted first to get some implication that it actually might work.
The experiments are by far no proof, but give a hope that it might work, so we can invest time and effort to build a better version and a better test stand.

Hey, the damper worked great! The vibrations really settle down after 5 hours. Ideally, the experiment should start only after this point.

Now we see that the current EMdrive is too weak to cause any effect. Two solutions for that: either decrease dampening, or increase the power. Ane because 5 hours of waiting is already quite long, you probably don't want to decrease dampening. The only way to go is more power.

Computers are very patient, so waiting more than 5 hours shouldn't be a problem.  In fact, I'd be very curious for a "null" test using reduced fin area and measuring how long until oscillations dampened.  The current damping level may exceed "forces" generated by 100mW.  Probably easier/cheaper to reduce damping than increase RF power.  As mentioned earlier, using ammonia gas filled cavity may provide an interesting possibility to amplifier RF power @ ~24GHz.

I´d like to try the ammonia but I´m concerned about danger to health.

How much power increase can you expect with a different electrolyte?

Ammonia gas has been used in MASER designs @ ~24GHz.  Filling the EM drive cavity with some amount of ammonia gas (concentration TBD) might amplify the RF.  https://hackaday.io/project/5596-em-drive/discussion-26483

Thanx. I am starting to build up the swimming platform now and will develop a PA with 500mW of power.
The PA will take some time, so I think we will have to perform the first waterfloating experiments with our actual oscillator.

Hey! I don't know if you've seen this, but over in the emDrive subreddit we've put together some test ideas and suggestions on how to reduce the noise for the baby EmDrive:

https://www.reddit.com/r/EmDrive/comments/3a3mmt/baby_emdrive_what_would_be_some_good_clear_tests/

Thanx a lot. I´ll check it out. We appreciate what you are doing and will try to put the ideas into our tests.

This is useless without negative controls and replicates.

In fact, it is worse than useless because all it does is create more noisy data that is impossible to interpret.

The measuring apparatus isn't as important as the experimental rigor! You can have a noisy apparatus and still get good signal if you are rigorous about using proper controls and are consistent among replicates.

At the very least, please do a negative control baseline and stop fiddling with the apparatus so different runs can be compared.

Why don´t you build a better one to prove us how to do it right?

Please don´t assume that we are stupid - try to make these tests by yourself than you will get some feeling for time, resources, money and Mr. Murphy. We are doing it for FUN and sharing everything - nothing is useless, except comments like that. 
Explanation: It´s a first prototype - we just want a HINT that it MIGHT work before we put time, money and work into a better setup.

Did you notice that some setup took only one or two days to build?
We want to discover things with ressources that we have and deliver some progress to everyone FOR FREE.

BTW some curves were pretty explicit, I´m sorry you don´t see that. Although I am not a statistician I can see with my naked eye what´s happening.

Of course you are right according to the test procedure - but the tone makes the music. Try it by yourself.

>Why don´t you build a better one to prove us how to do it right?

Because we want to see if this works first. Science is incremental, so first we must learn what we can from your experiments. To do that we need to see a series of consistent measurements with your device compared to a baseline so that we, as a community can learn.

Sorry if this comes off as harsh, but that's research. Better to say the hard truths quickly and to the point. Otherwise, this whole em-drive thing is going to go the way of cold fusion.

You've put a lot of hard work into making the device and it was REALLY frustrating to keep seeing scattered noisy results. So I had to say something.

>Explanation: It´s a first prototype - we just want a HINT that it MIGHT work before we put time, money and work into a better setup.

It doesn't matter what it is, if you can't put data into context, it may as well be a banana on wheels. 

Accuracy is useless without consistency. If you keep changing the apparatus and configuration, you'll never find that HINT you are looking for. 

>Did you notice that some setup took only one or two days to build?

Does that somehow change your ability to repeatedly measure something?

Ad hoc apparatuses are common in science. It is easy to account for error caused by an inaccurate apparatus as long as that inaccuracy is quantified by a baseline.

Good science can be done with duct tape and balsa wood. What matters is thoughtful experimentation.

>We want to discover things with ressources that we have and deliver some progress to everyone FOR FREE.

Cool. Great. Amazing. Wonderful. I'm here to help, and that means not contributing to a circlejerk.

>BTW some curves were pretty explicit, I´m sorry you don´t see that. 

I very much disagree.  Nothing presented so far has been at all convincing. Remember, EmDrive thrust is a very extraordinary claim, particularly with a device this small. It requires robust repeated evidence.

This is why I am being harsh with you, because I want to see it work. 

>Although I am not a statistician I can see with my naked eye what´s happening.

*siren* Bias Alert *siren* 

Sorry, but no you can't, and if you keep that attitude no one will believe your results. 

Congratulations on the damping the noise!  Seems like setup may now be over-damped.  Improving tangential orientation of cavity, altering/adjusting damping techniques (fin area, oil vs water, isolation, string length, etc), and waiting until oscillations subside before RF activation provide lots of variables to work with.  Does your RF source allow use of PWM for a modulated "on" period?  Some have postulated that charge/discharge cycles might increase "forces".  Good luck!

Sure, too many variables changed - this measuring technique is suboptimal.
We are working on a swimming platform now. Until it´s ready we´ll just continue some experiments. The test rig is already there, and any information is valuable.

Ein missverstandnis. Dampfungs gut, aber jetzt zu viel? Viele variablen zu optimieren. :) Viel Gluck!

What was wrong with the magnetic levitation test? Any plans on returning to that process?

from http://www.reddit.com/r/EmDrive/comments/3a3rce/had_an_informal_interview_with_movax_baby_emdrive/

Q: Why did you give up the levitating EMdrive? It seemed promising, with frictionless mounting and all.

A: The levitation device is overwhelmed with the weight. It's just a toy that normally supports a spinning globe, and we put about 20g more into it than originally. The principle is quite solid - it uses a Hall sensor to detect the current position of the magnet, not one of these dumb light switches. The main feedback loop works at 40kHz; that should be quick enough. But in our setup, the whole thing wobbled up and down constantly - it's impossible to get the weight distributed correctly. By the way, that's the reason for the strong fall-off at the end of the graphs: the device is always a bit lopsided, and when it rotates slow enough, it "falls" into a stable orientation and doesn't have enough momentum to start another rotation. Remember, this thing only turns once every ten seconds or so.

Okay, this is clearly not working as a reliable method of testing for you guys.  But, do you have another plan? Use oil if you continue with this method.  Apparently, that would help a lot according to Rodal.

Yes. We will put it on a swimming platform, stabilized by a ring magnet on the ground an another one under the platform.

Actually, the only thing that changed since last time is the amount of dampening. The data after x=900 is probably what we want to use as a solid basis for experimentation. The only issue is that setup for the "forward" direction wasn't dampened as well as this orientation.