I did other measurements in different conditions trying to understand where the noise comes from.

I plot the raw data AC (lockin Y vs lockin X) divided by the DC.

Plot1: Tamasize sample sit on blocks (already plotted in last report) makes more noise than the small sample, even dividing by the DC.

Plot2: Noise level (standard deviation) in two different days have a repeatability of about 10%.

Plot3: Both samples tightly fixed at the board using the new mirror mounts. Small sample makes more noise than the tamasize sample. I really dont understand this. Also the distribution for the small sample is strange, more points on the tails of the distribution.

 

I moved the detection unit 35mm further to check if the noise depends on the distance of the detection unit.

The detection unit is made by the focusing lens (f = 50mm), the smal reflecting sphere (f = 1.25mm) and the detector.

 

Plot4: at a further position the noise is less (I expected the opposite)

Plot5: without dividing by the DC the ratio between the standard deviations changes a little bit.

 

 

I placed the detection unit at an intermediate distance: 25mm further than the original position.

 

Plot6:Comparison of small and tamasize sample. The small sample makes more noise the tamasize mirror.

 

Plot7: It looks like the chopper vibrations contributes for 25% on the total noise.

 

I used the internal oscillator of the lockin amp to avoid any phase noise due to the chopper frequency instability.

Plot8: Tamasize sample. The phase noise doesn't look to have significant  contribution.

Plot9: Comparison of Tamasize and small sample. The reference signal is the lockin internal oscillator. Situation is same as with the chopper reference.

 

Result. Most of the noise doesn't come from the chopper vibration.  I still don't understand where it comes from.