Per here the main thing wrong with the interferometer is the vibrations, no surprises there. I recall seeing a demo at photonics west which in retrospect was probably minus k technologies. It does vibration isolation with a flexure element that has negative stiffness. If you knew what to build, it would probably be pretty easy.
Simulation of buckling loads
Here is claudes first attempt at simulating a buckling load:
The middle plot is supposed to look like this:
From here. It kiinda looks like that but not really.
Real isolation
I have been furnished with a prototype vibration isolation mount based on the above negative stiffness mount.
Setup
Sadly the whole rig is just a bit too heavy, and so it bottoms out the setup:
So it’s not in the flat part of the force-displacement curve.
Results
It does something, but still pretty bad. But as mentioned it’s not being operated at the right point.
Proper balancing.
You can see from the above pics that there are 4 1/2” steel rods holding everything up. I removed these and replaced it with 3d printed 1/2” rods to cut down on the weight. This was enough to (just) bring it below the operating points. On thorlabs mirror stage, one 4mm allen key, and one 1/4”-20 bolt on top of that puts it in the sweet spot.
Vibration isolated - Spectrum
Vibration isolated - time series
This makes the difference abundantly clear:
Power supply sensitivity
Here is a plot of me twisting the knob on the power supply back and forth to modulate the current going into the laser:
Considering I was swinging the power back and forth by quite a bit, I would estimate many 10s of mA, this actually points away from the residual noise being power supply related I think since that implies that the benchtop power supply is implausibly noisy.
Vibration correlation
If the remaining noise in the interferometer output was due to vibration, then perhaps there would be a correlation between the voltage across a piezo mic and the output of the interferometer. You might be able to see this in the xy mode of a scope:
…or, perhaps not. This isn’t a definitive proof though. I don’t think that the piezo voltage is linearly proportional to the displacement of the system
Pic of setup:
