you should try this very simple, 10-minutes-to-setup test, with the elements you already have (pic of the test here: http://imgur.com/vkeHVhb). the whole point should be killing that huge noise you are having. you have that simple, cordless lightweight system that you can use to design several kinds of tests, but you need to focus in stoping the noise. put it in a container, floating in the water, and make sure it settles down and stops moving. then leave it that way for say 10-15 minutes, it mustnt touch the walls nor move. then power it and leave it on as long as you can or until it starts moving :) but dont bother turning it on if you cant make it stay still when its off. BTW keep going for it and good luck!

Great, we will make a setup like this in the next test. Goocy proposed to put a magnet on the ground of the tank and another one on the rig so it doesnt float away but can turn freely

That would be great to keep the rig in place. Are you planning on adding power with another battery to increase the thrust?

here is a comment of someone with an idea of a cheap way to dampen vibrations 

http://www.reddit.com/r/EmDrive/comments/3a3rce/had_an_informal_interview_with_movax_baby_emdrive/cs99yc8

Can you tell us the ranges exactly when the thruster is on?

Can you confirm what minutes x 3 means. the data goes to 1493, does this mean it ran for 4479 minutes? Or 497.6 minutes.

Here is a preliminary analysis like the one I did for the previous data set:

1400 <-> 1400/3 minutes

The timelapse samples every 2 seconds to a 30fps video. The x-axis is videoframe/10

Any chance the next setup will have the cavity closer to perpendicular with respect to torsion axis?  Improving the balance beam (and cavity) orientation should help reduce some of the potential noise sources and help focus any available "thrust"/force into a measurable direction.

Very exciting test!

From the last image it seems that some of the wires are not twisted. This will make your test rig act as a compass needle where the torque depends on the current. If you have wires that carries a lot of current they should therefore be twisted to avoid disturbances.

I have tried to filter the data a bit:

Great, thank you for that. The current is about 60mA.

The next setup will be a lot more clean. As I ran this test already in this configuration, I won´t change the wires. I will just turn the cavity around and see if the deviation inverts.
If the magnetic forces occur, they should occur in the same direction, and they seem to be even lower than the thrust.

Yes, the next test will show if the magnetic field is a problem if you
make sure that the orientation of the rig is the same in both tests. I
will process the new data when the results are in :-)