Paul, do you have a airco in your testplace? My experience is that a controlled temperature increases the quality of the test result. If its warmer, you get more unwanted fringes.. (fata morgana effects) 

I don´t have one, but the temperatures are mild here at this time. I´ll cover the complete setup in a box to avoid larger airflows. However, the RF source is heating away over 1 Watt which might result in some disturbances.

attach a cooling ribbon onto it (from a microprocessor perhaps) You need to dissipate the heat. You could also place a metal panel between the experiment and the power source..

The power source will be on the platform together with the EMdrive, so the displacement of the whole EMdrive setup will be measured. The RF housing will have a radiator so the heat can go upwards and hopefully will not make too much noise

How do you stabilise the interferometer? I used to make holograms and i know that any vibration on the ground will be visible on the endresult. This  system is no exeption! I used  soft rubber buffers and a stone slab for weight to reduce any vibration from the device AND the enviroment.  even breathing and heat diffrences can falsify the results.

When I do nothing, the pattern stays stable. But I fixed the second mirror - in the final setup the mirror will be on the EMdrive which will be beared elastically, so I expect vibrations too.
When I knock with my finger on the chair standing nearby the table, there is heavy movement detectable though.
I used a 12mm aluminium plate as base, maybe that helps to keep it stable. I live close to countryside and in the evening there is almost no traffic, I assume that´s also a reason why it stays stable.

Update: There are changes, but they are very slow, like 10 minutes for a half wavelength, sometimes shorter, sometimes longer.

try to put small disks of memory foam under the table legs.

Wow, that looks awesome, and very professional!  Nice work, can't wait to see your code, and a video, that was fast! What kind of laser did you end up using?  And camera?

I used a 10$ red laser module which showed very stable. But I broke it due to rotating it to get equal bright beams through the beam splitter cube. My friend Jack got me a replacement laser within half an hour (respect to that) so I could use the weekend to work on the software. I ordered two more, just in case. Good interference patterns occur with a laser  whose point is elongated. You can put a biconcave lens in the last path so the screen can be put close to the cube.

I use a playstation PSeye camera. The reason is that I used it in other projects before and therefore have the toolchain for it. I can scan 60 frames per second which should be enough for slowl displacements.
The current resolution I get is about 10nm - at night time where there´s no traffic. Else there is some noise. But it´s nice to see that electrical EMI is no issue at all :-D Thanx again for that hint.