Waht I did: estimated Q factor after attached coil magnet actuator.

I estimated Q factor of Roberts linkage after attached coil magent actuator.
This Robert linkages' resonant frequency is 0.67Hz.
When I estimated Q factor, I used the ring down curve from 2000s to 2500s(just Fig 2).
Estimated Q factor of Roberts linkages is 3.91×10^3.

Blue points are mesurement and red points are fitting.
Vertical axis is read out of photo seosor that can detect displacement, horizontal axis is time.
Fig 1 is over view of the ring down curve.
Fig 2 is the data that is used for estimating Q factor(2000s to 2500s).
Fig 3 is over view and fitting results.
Fig 4 is the ring down curve that is used for estimating Q factor and fitting results.
Fig 5 is also the ring down curve and fitting results from 2000s to 2020s.

What I did: measure the transfer function from coil magnet actuator to photo sensor.

I measured the Roberts Linkage's transfer function from coil magnet actuator to photo sensor.
First of all, I attached magnets on the suspended mass, and also set a coil.
Fig 1, 2, 3, 4 is a setup of coil magnet actuator.

When I measured the transfer function, the data was measured from 0.1Hz to 10Hz and also measured form 10Hz to 0.1Hz.
The reason is that a resonant osillation of Roberts linkage remained for a while, so I can't measured the trasfer function appropriately after passing throught the resonant frequency.

Fig 5 is the transfer function when I measured it from 0.1Hz to 10Hz.
Fig 6 is the transfer function when I measured it from 10Hz to 0.1Hz.
Fig 7 is combined one. 

[Marc, Shalika]

As our calibration and vi are now finalized we are now setting up our final calibration before starting birefringence measurement.

We made a black cover box from 2 un-used optical lever covers. We drilled one small hole for the laser input and another one for the several cables we need.

The ambient light from the power meter or the camera were reduced by a factor 100.

[Marc, Shalika]

Following the speed improvement of the LC voltage control we implemented a sine modulation of the voltage.

However, we found that the resulting retardance is different if the voltage is increasing or decreasing.

We decreased a lot both the sine freq and sampling freq and could resolve this issue.

This seems related to the LC different switching frequency when applying increasing or decreasing voltage with decreasing voltage being faster.

To mitigate this effect, we also implemented a decreasing sawtooth function and plan to mainly use this one for our future calibration and measurements.

I checked the voltage and curent in the SIPs today.

[Shalika, Marc]

Overview: The speed of saving data/characterization is 80 Hz. 

Details:

We changed a "lot" of stuff. Techniquely we were removing everything one by one and seeing how they affected the speed. We did this with every single component in our VI. We were simultaneously optimizing speed by removing VIs which were not so important. Previously our speed was 8Hz, so we were acquiring 8 points per second. Now we have around 80 points per second, i.e 80Hz. 

Refer to Fig 1 for more details, but below are the most essential parts which helped optimize the speed. 

1. Temperature controller was extremely heavy. It doubled the speed when we removed it. We brought the control outside the main loop. 

2. We did the same with the Power meter and Polarization camera. 

3. And, we are now using global variables to access and save data in the main loop.  

I started the SIPs between the EM2 and the mid point in the south arm. The power supply #3 (DIGITEL 1500) drived "N-S P5" and "N-S P6", and the power suply #4 (DIGITEL MPC) drived "N-S P7". Applied voltag and current were changed as follows.

[Just after starting]

[After 3 hours for P5&P6 or 1.3 hours for P7]

I checked the voltage and curent in the SIPs today.

I started the SIPs between the NM2 and the mid point in the south arm. The power supply #1 (DIGITEL MPC) drived "N-S P1" and "N-S P2", and the power suply #2 (DIGITEL 1500) drived "N-S P3"(photo) and "N-S P4". Applied voltag and current were changed as follows.

[Just after starting]

 [Ater 2 hours]

Yohei,

What I did:

I have measured the beam profiles of the Mephisto laser. I have used a f250 lens to focus the beam, then estimated its original q-parameters of the laser.

Result:

The measured beam parameters (Fig1).

* z=0 is set to 100 mm away from the laser head. See the yellow tape in the attached picture (Fig2).

The estimated original parameters of the laser.

The weist is 60-70 mm inside the laser head. The weist position is supposed to be 90 mm inside the laser head, but the estimated values suggest it's almost near the laser head. I don't know why it happens.

** Sorry but there is some error on the first figure.

What I did: estimated Q factor

I tryed to measure Q factor of Roberts Linkages that resonant frequency is 0.67Hz by ring down curve.
When I estimated Q factor, I used the data from 500s to 1400s (just Fig 2).
Estimated Q factor is 5.97×10^3.

Blue points are measurement and red points are fitting.
Vertical axis is read out of photo sensor that can detect displacement, horizontal axis is time.
Fig 1 is over view of the ring down curve.
Fig 2 is the data that is used for estimating Q factor.
Fig 3 is over view and fitting results.
Fig 4 is the ring down curve that is used for estimating Q factor and fitting results.
Fig 5 is also the ring down curve and fitting results from 600s to 620s.

Yohei,

I started to built a new set up of the speed meter experiment. 

What I did:

I put a QWP and HWP in front of the source laser (Mephisto, Innolight), mazimizing p (holizontal) polarization.

Then, I put a Faraday Isolator (FI), optimizing its extinction ratio.

I measured the transmissivity of the FI:

Input = 10.3 mW,

Ouput = 9.3 mW,

Transmissivity = 90.3 %.

When measuring the beam power by a power meter, I attatched a ND filter, ND=2.0. From this value, the source lase power can be extimated as ~ 1W, keeping its original value.

I evacuated the duct with the large RP so as to be lower than 0.1mbar. After that, the GVs between the pumps and the duct were opened except for the south end.

[Marc, Michael, Yuhang]

We measured FIS with various green power as reported in figure 1.

However, one issue was that CC1 and CC2 error signal were extremely glitchy with every few seconds a huge increase in their levels.

Note that to speed up the FIS recovery we did not tune too carefully the various servo gain.

We will do it soon and also investigate PLL phase noise.

We suspect this is the reason why we can not see improvement on the phase-noise (fig2).

Today, all (four) pump units along the south arm have failed. I supporsed that the dry-pump replaced yesterday in the south end faild at first, then other pumps also went to stop. The pressure of the arm duct increased to 20mbar. The dry-pump in the south end was broken with "ALM05 MP STEP", The other pump units were recovered. Since the pressure  of the arm duct is too high, the GVs between the pumps and the duct are closed. It is necessary to evacuate the duct with the large RP so as to be lower than 0.1mbar.

What I did: I fited the transfer function of Roberts linkages, and estimated the resonant frequency and Q factor of it.

I analyzed the transfer function of Roberts linkages, and try to estimate the resonant frequnecy and Q factor of it.

When I estimated the resonant frequency and Q factor, I do it separately. The reason are as follows.
First, When I measured the transfer function, the gain around the resonant frequency is crashed by out of the linar range of the photo seosor. So I need to omited these bandwidth. Meanwhile I need the bandwidth of resonant frequency when I try to analyze Q factor.
From the above situation, I estimated the resonant frequency and Q factor separately.

The result were as follows. The pictures were also attached.
 

Position of center of mass means lenght from suspension point to center of mass vertically. Negative means that center of mass is below suspension point.
Estimated Q factor are so big and strange, and I thought It was caused by not enough time resolution.

What I will do: I measrue Q factor by ring down curve fitting.

[Marc, Michael, Yuhang]

First we realigned p-pol, BAB and green into OPO.

For BAB, we installed a power-meter before the homodyne.

We restarted the homodyne power supply (-19V, 0.06A) and confirmed the usual BAB transmission from OPO (~400 uW or ~200mV on homodyne sub DC).

We balance the homodyne using LO beam.

We realigned LO and BAB into AMC.

Then, we checked the OPO non-linear gain with MZ offset 4.2V.

We found T = 7.122 kOhm and p-pol freq = 180 MHz was maximizing amplification.

We measured with green (26mW) maximum of BAB of 1.04V while wihtout green it was 0.186V meaning a non-linear gain of about 5.4.

We lock CC1 and CC2 using the DDS3 config saved as '20230406_dds3'. Note that we found out that DAC1 and DAC3 cables of DDS3 are swapped. We should remember to modify accordingly the DDS3 config after reconnecting properly the cables.

Using the new SR785 we measure squeezing, anti-squeezing, shot-noise, homodyne noise and SR785 noise as attached in figure 1.

Note that the data are saved in .78D. To convert them into text file we have to save the data into the folder ./python/conversion and then use the command 'SRTRANS /Oasc /D SHOTN2.78D SHOTN2.txt' where SHOTN2 is the filename to convert.

Because we had to 'randomly' tune various gains to speed up debugging, we have some noise structure about 4 kHz. We will remeasure the OLTF to better tune their gains.

In any case, we measured about 5.9dB squeezing and 11.5 dB anti-squeezing !

This is compatible with our loss estimate in the wiki. We will measure squeezing again to see the EOM replacement effect on phase noise.