NAOJ GW Elog Logbook 3.2
[Aritomi, Yuhang]
1. Achieved lock of SHG, GRMC and MZ. The power going to OPO now is 45mW.
2. BAB beam power now is 145mW.
[Aritomi, Eleonora, Yuhang]
Since we found the problem of unbalance of homodyne, we replaced the wrong coating lens.
After that we measured the noise spectrum. We can see from this spectrum that
1. The shot noise is limiting above 900Hz. However, there are several peaks still existing in the shot noise region.
2. There is a noise with slope of 1/sqrt(f) below 300Hz. Noise source could be scattering, beam jittering, vibration.
3. We recognize there is 600Hz which comes from molecular pump for vacuum.
[Aritomi, Eleonora]
Today we tested the possibility to operate the local controls of the end mirror, by shipping the error and the correction signals with the optical fibers system in place in TAMA.
Details of the optical fiber system are reported in entry #1068. We used the boards "A" (4 input and 4 output channel). See pic 1. We sent two error signals (pitch and yaw) toward the central building and we recived four signals (corresponding to the 4 coils).
In order to perform the test, the signals have been connected to the ADC and DAC of the input mirror. We also used the input mirror CPU and labview VI to test the control. (We temporarily copied the filter and the driving matrix values used for the end mirror)
Conclusion: we could succesfully close the local control loops in this configuration which means that we can go for a stand-alone control system (with all the DAC and ADC in the central area), as planned.
It seems that with the new year, the labview PC is not able to deal with Japanese characters anymore.
Because of this, we had some troubles to run the local control labview project as the path of the files contained some katakana.
The problem was solved by changing the name of the desktop folder from katakana to romaji.
(The origin of the problem is mysterious since nothing has been changed from the last time we used the PC.)
[Aritomi, Yuhang]
First we changed the position of laser illumination point by moving lens in three directions. However, the difference is still the same as before.
Then we measured some values as following.
| Total power | PD1 power(before lens) | PD2 power(before lens) | modulation on PD1 | modulation on PD2 |
| 1.052mW | 489uW | 504uW | 608mV | 752mV |
The ratio of power is reasonable. But the ratio of modulation is not. So we guess the problem should between lens and oscilloscope.
The reason can be: incident angle resulting in different responsivity; lens problem; PD problem
We first tried to change the angle of incident beam on PD1 by doing the 'alignment' of BS and lens. Here alignment means makeing the modulation signal on PD small by BS and recover by lens. However, it doesn't make any difference.
Then we think PD should be fine. And the last check is for lens. However, there is no space to put power meter after the lens. So we take it off from the original position. And put it before the IRMC. The power before and after this lens is
| before | after |
| 1.39mW | 1.09mW |
The ratio of these two power is 0.784. If we make this ratio as 1, the modulation on PD1 should be 775mV. It will be almost the same with PD2. So it explains why we have so large unbalance in homodyne.
This lens is from Nowport lens box(for infrared), however, it seems a green coating.
Conclusion: the unbalance is from using of wrong lens. We have already ordered a proper lens. It should arrive soon.
[Eleonora, Yuhang]
Here I put the spectrum containing all the noise source we have for now. The measurement is done after a reasonable common mode noise rejection. The shot noise level we are using here is -135dBVrms/rtHz and it's from Henning(corresponding to 1mW of incident laser power).
[Yuhang, Eleonora]
We made a tentative design of the position and dimensions of the breadboard we plan to use for the AA quadrants.
It is shown in the first attached scheme.
The dimension we selected is 300x450 mm (https://www.thorlabs.com/thorproduct.cfm?partnumber=MB3045/M)
According to this design, the reflected beam collected by the Farady isolator is sent to the board by a steering mirror and reach the board after a path of 637mm. (See second attached scheme.) It might be possible to install a lens on this path.
The dimension and position (above the AUX lasers) have been chosen to allow the access to all the optics. If a larger breadbord is needed we can cosider to extend it above the "PLL area".
Some pictures of the bench are also attached. Here we used a plastic sheet of the same sized of the board to better visualize the occupied space. Fixing the breadbord posts will require some adjustment of the cables but seems feasable.
[Aritomi, Yuhang]
This is work on last Friday.
After winter holiday, the second faraday isolator seems to be broken.
Figure 1 shows the transmission of the second faraday isolator.
The beam shape is strange though the beam seems to go through the faraday isolator.
So we removed the isolator and a half wave plate right after it as shown in figure 2.
After re-alignment of SHG, mode matching of SHG is 1.5/(1.5+0.08)= 95% as shown in Figure 3.
We plan to buy a new isolator.
[Aritomi, Yuhang]
According to Henning result, the shot noise should be -132dBV for 2mW of incident light. In our case, the machine noise is -100dBV. So we use SR560 to amplify it by a factor of 100 means 40dB. So the shot noise level should be -92dBV.
First, we measured the noise spectrum when one of the PD is blocked. The result is shown in the attached figure one. It is clear that not even part of noise spectrum is dominated by shot noise. Even at highest frequency, noise level is around -88dBV.
Then we decide to change pitch of steering mirror to change power on the second PD. Maybe this can bring beam away from optimal position. Or the change of beam direction reduce detection efficiency. In the end, we reduced common noise rejection from -30dBV to -74dBV, means we reduce the modulated signal by a factor of 100. The result is shown in the attached figure 2.
Finally, we turned off modulation. And then we look at the signal from homodyne on spectrum analyzer. The result is shown in the attached figure 3. Then we improve the locking condition of IRMC, make the locking point as close to peak as possible. Then the signal becomes figure 4. This is reasonable because around peak the amplitude change is less. Also seems the noise comes from amplitude noise.
To do list:
1.We can also change the pitch of BS to see if we can increase signal on the first PD.
2. Try to improve locking performance.
3. To see difference when noise eater is engaged and not.
4. Noise hunting
[Aritomi, Yuhang]
The procedure we did for aligning homodyne is:
1. Lock IMC and driving IMC end mirror with a ramp signal(1kHz 50mV). This will drive the output of IMC moving around the locking point. If we lock well on the peak value of IMC transmission. We will see the double frequency of the end mirror driving frequency. In this case we can use the modulated IMC transmission signal.
2. Connect homo-dyne and put it roughly in a good height and good horizental position. Here look at modulated IMC tra signal and make it maximum. Then fix homodyne.
3. Put lens in roughly the corresponding focal length of lens away from homodyne. Then roughly make the signal maximum again. Then fix lens.
4. Adjust lens so that we can further maximize.
After that we take the modulated signal both on oscilloscope and spetrum analyzer. Also take the signal when one of PD blocked or neither of them blocked.
The first three attached figure is for oscilloscope. The first one is when the second PD is blocked. The second figure is when the first PD is blocked. The third figure is when neither if them is blocked. (first PD is the nearest PD to IMC) (note here IMC is not locked on peak)
The next three figure is for block second PD, block first PD and no block. (here IMC is locked on peak)
Optimization is still needed.
[Aritomi, Yuhang]
Since we receive message from Henning, the voltage from pin 1 to 3 to 5 should be from -19 to 0 to 19V. So it means we should swap the connector. After do that we can see the signal on oscilloscope and its value is reasonable. Also the current coming out from power supply becomes not zero. As shown in the last attached figure. So we can say homodyne works well now.
After that, I measured again the dark noise of homo-dyne.
Firstly, I measured the electronic noise while SR560 is set gain as 1000(means 60dB). The result is shown in the first attached figure.
Then I measured the signal from homo-dyne whlie no light going inside homo-dyne. The result indicates the dark noise of homo-dyne is -93.2-60 = -153.2 dBV/rtHz. The result from Henning is -156.2dBVrms/rtHz(shown in attached figure 3). And since there is relationship Vrms = V/sqrt(2). This factor sqrt(2) is exactly 3dB difference. This means our measurement perfectly matches Henning's measurement.
The first two pictures shows the set-up of our power supply. As you can see from the picture, the voltage we are giving is +/- 19V.
The third picture shows the connection of the cable to the voltage supply. The connection follows the content of labels.
Finally, as pointed by Henning, I checked the voltage between pin 1, pin 5 and pin 3. These are shown in the attached picture 4 and 5. The voltage between pin 1 and 3 is 19V. While the voltage between pin 5 and 3 is -19V.
This result means the voltage going to homodyne should be fine.
[Aritomi, Eleonora, Matteo, Yuhang]
Investigations carried on yesterday and today seem to confirm that we might have a problem in the powering of the homodyne detector.
The main clue is that we cannot observe any change in its output signal when the power cable is connected and when it is not.
We do see a change in the spectrum of the signal when we shine a laser with an amplitude modulation (see Fig. 1) but the signal is the same even if we disconnect the power cable.
We checked the power cable connections and confirmed that the voltage arriving to the homodyne through it is the correct one.
Fig. 2, 3 show the homodyne detector and the setup.
All the electronics in the cleanbooth have been switched off, except for the LabView supervisor PC.
The oplev lasers, the coildrivers and the target labview PCs are also off, both in the central and in the end room.
All the used cleansuits, hoods and shoes have been sent to laundry.
Participaint: Aritomi, Eleonora, Matteo and Yuhang
First we checked the homodyne dark noise. The purpose is to compare this result with Henning's result. We are using network analyzer, but itself's noise level is not low enough to see the dark noise of homo-dyne. So we decided to use Stanford research SR560 to amplify the homodyne noise. We gave it a factor of 100(40dB). Then we got the result as shown in attached figure 1 and 2. The difference is they have a marker located at 1kHz and 10kHz, others are the same. We can see below 3kHz, the noise should be dominated by electronic noise. And above 3kHz, the noise level should be the noise level of homodyne. In this case, the real noise level of homodyne should be around -122-40 = -162dBV/sqrt(Hz). However, the result from Henning is
-156dBV/sqrt(Hz). This is not a neligible difference.
We are arriving to the point to operate homo-dyne. We power up homo-dyne with a DC voltage supply at +/- 19V. For checking both the alignment of the beam inside homodyne and also if the homodyne work well, we send an amplitude modulation to the main laser beam. At the beginning, we send a 100mV pk-pk and 1kHz signal into the current control of main laser. However, this modulation seems not stable. Then we decide to modulate the PZT of IRMC, in this case, we moved the locking point of IRMC around the TEM00 peak. Then the modulation became very clear. Here the clear or stable means if the signal we see on the oscilloscope shows clear line or can be triggered.
We also find out that the the even we send the voltage +/- 19V to homodyne. The current going inside seems like zero.
And when we see the 1kHz amplitude modulation signal on the monitor PD with amplitude of several volts. We can only see the same signal on homodyne with amplitude of several millivolts. Seems like we didn't switch on homodyne.
The last figure shows the homodyne on our bench.
Participaint: Aritomi, Eleonora and Yuhang
According to the optical layout, the lens should be 250mm before beam waist. And the beam waist is 390um. By putting a 30mm lens, we found the beam waist after lens will be 22um. The distance of the waist from the lens is 29mm. The corresponding rayleigh range is 1.4mm. The aperature of homodyne detector is 0.5mm, so we should make the going inside beam smaller than 100um. The range of beam smaller than 100um is 12mm.
The measurement result agrees with the simulation. The result is attached in the figure2.
The attached figure and PDF is the same. They show the space we may have the leg for breadboard. The space in this case is roughly 90cm*75cm. It covers all the space for PLL and two auxiliary laser heads.
1. The p-pol beam is very sensitive to the mirror mount. Even you touch the mirror mount, the alignmet condition will be changed.
2. The OPO's BAB transmission is also very sensitive to mirror mount.
3. The PLL locking is not robust enough. It can only last for several tens of minutes. And very diffcult to acquire, we can close and open software again to make this better.
4. The second Faraday's mount keeps moving. And its transmission has a lot of scattering.
Participaint: Eleonora and Yuhang
We convert BAB into p-pol and then measured the beat between it and the LO at the homodybne BS.
The result is shown in the attached picture. The visibility is (Vmax-Vmin)/(Vmax+Vmin) = (4.26-1.26)/(4.26+1.26) = 0.5435
The BAB power is 0.125mW and LO power is 1.24mW. The expected visibility is 2*sqrt(P1*P2)/(P1+P2) = 0.5768
So from these values, we should have 0.5435/0.5768 = 94.22% of matching of two beams. This is complaint with the matching we measured before.
There is one thing block us from measuring visibility, it is the use of lenses. Before we used lense, we can see the beam is flashing but cannot see this signal on oscilloscope.
By considering the level of visibility, I did the simulation of degradation. As shown in the attached figure two. Now because of only the reason of visibility, we degrade squeezing by 2.74dB.(I assume here the initial squeezing level is -10dB, the degradation is different for different initial squeezing level)
