Given the fact that a thick sample changes the optical path of the probe, I wanted to see how and why the noise level changes when I change the position of the detection unit. The detection unit is made by one flat mirror at 45°, a f=50mm lens, a reflecting sphere f=2.5mm, and the photodetector.
I turned OFF the chopper to avoid any possible vibration, I set the lock-in internal oscillator as reference frequency (demodulation) at 420Hz.
I connected the oscilloscope at the photodetector (to see the DC signal), and at the output of the lock-in amp (to see the AC signal *1e6). I took some quick measurements, for different positions of the detection unit. The DC has a repeatability of 0.2V, the AC measurement is very rough, just an average of the signal in 10s. Every time I moved the detection unit I had to realign the beam on the PD, tuning the position of the 50mm lens to maximize the DC. The position can be changed only by 35mm, the length of the micrometer screw.
In the following table, there are the DC and AC values at different positions and for different samples. A higher position value means the detection unit is closer to the sample. Hence, 0mm is the furthest point where I could place the detection unit. To move it more it's necessary to unscrew the unit from the board.
I used three samples: the Sapphire small sample diam 1.5" x 5mm, the Sapphire Tama-sized sample diam100mm x 60mm, and the glass KAGRA-sized sample.
| No sample | Small sample | Tama-size sample | KAGRA size sample | |||||||||
| Position | DC | AC | DC | AC | DC | AC | DC | AC | ||||
| (mm) | (V) | (mV) | (V) | (mV) | (V) | (mV) | (V) | (mV) | ||||
| 34 | 7.4 | ~100 | 6.9 | ~140 | 8.5 | ~500 | 8.6 | ~1100 | ||||
| 30 | 6.8 | 6.4 | ~100 | 8.5 | ~500 | 8.6 | ~900 | |||||
| 25 | 5.8 | ~80 | 5.4 | ~110 | 8.5 | ~400 | 8.6 | ~800 | ||||
| 20 | 4.9 | 4.6 | ~90 | 8.0 | ~150 | 8.6 | ~900 | |||||
| 15 | 4.4 | ~80 | 4.0 | ~100 | 7.4 | ~120 | 8.6 | ~900 | ||||
| 10 | 4.0 | 3.6 | ~90 | 6.6 | ~115 | 8.5 | ~600 | |||||
| 5 | 4.4* | 3.6 | ~80 | 5.9 | ~80 | 8.5 | ~400 | |||||
Looking at those data, I can say:
- the DC decreases when the unit is placed further. This is reasonable considering the finite size of the PD and the divergence of the beam.
- In the case without any sample, when the unit position changes, the AC noise level doesn't change a lot.
- In the case with the small sample, when the unit distance increases, the AC noise level does decrease, maybe proportionally to the DC.
- In the case with the Tama-size sample, when the unit distance increases, the AC noise level changes a lot, compared to the DC.
- In the case with the KAGRA-size sample, when the unit distance increases, the DC is pretty constant and the AC noise level changes a bit.
The hypothesis I have in mind is that the probe spot size makes an important role when compared to the detector size.
I will use my simulations to try to reproduce the behavior shown in those measurements and try to find an explanation.
The signal that I get with the oscilloscope comes from Lock-in CH1 OUTPUT. It is much higher than the signal recorded by the vi (the AC signal in entry 252, for example).
I read the sr830 lock-in manual and I found that the CH1 OUTPUT voltage is proportional to the AC signal according to the following formula:
Output = (signal/sensitivity - offset) x Expand x 10 V
The Expand factor is 1, the sensitivity is 1mV, as we can see in the picture of the front panel.
So in the case of signal = 15uV , for example, I get Output = 150mV, a factor 10^4 higher