NAOJ GW Elog Logbook 3.2
Thanks to thermistor data sheet from yuefan and the data tomura-san and I took(see e-log 627). I finished the SHG characterization work.
Since the data sheet doesn't give us a formula but only a few points, I use them to make a cubic interpolate. Then I use the interpolation data to make the plot so that you can see the relationship between crystal temperature and Laser power.
The NTC used seems to be SEMITEC 103JT-050.
According to the datasheet, the R25 = 10kΩ ± 1% and B = 3435K ± 1% (note that this value is correct only between 25deg to 85deg Celsius).
Today I finished the OptoCad simulation with practical scenario. I used the real measurement distance for optical components on the bench, and the in-vacuum distance from yuefan's e-log entry 441. I tried also to fine tune(means to move it with like two milimeters) the position of the last infrared lens on the bench. However, the results turn out to be similar with each other when you just look the simulation output picture. And also you can find the output data is similar. For example, the beam size around FI for them are 1.06mm and 1.36mm individualy. See attached file1~4.
However, we can see from the above attached simulation pdf file that the entering beam of PR chamber has some anomalous behaviours. You can see from attached file5, it shows only the incident beam and it is really good. That means bad thing is reflection. I also checked the output txt file and confirmed the anomalous comes from the reflection of cavity. However, since now I haven't found that why it is like this. I would like to talk with Matteo next week.
The usual procedure to align the beam for a best absorption signal is:
- to make the 2 beams pass through the pinhole
- maximize the DC centering the probe on the detector
- maximize the AC aligning the pump.
Adding the second probe beam makes things a bit more difficult because if we move the pump, we lose the alignment with the other probe. So, the maximization of the 1310nm probe AC signal have to be done without moving the pump.
Since the 1310nm probe turns out to be difficult, we decided to follow the usual procedure for the 1310nm laser, moving the pump to maximize it. Then try to recover the alignment on the red probe without moving the pump.
38.85mm is the position of the maximum of the scan with the red probe. Then we move the pump to find the maximum with the 1310nm probe.
filename:
Mon, Jan 22, 2018 3-01-33 PM.txt ; pd distance 83mm. Stage pos: 9mm
Mon, Jan 22, 2018 3-53-55 PM.txt ; pd distance 95mm (not better)
Mon, Jan 22, 2018 3-59-17 PM.txt ; pd distance 65mm
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
After trying the alignment many times, I noticed that the 1310nm probe is not focused properly, so we decided to measure the beam profile.
We discovered that the waist position is about 6 cm after the crossing point.
So we decided to change the focusing lens from f200mm to f300mm and to move the lens backward.
Yesterday we have worked on the characterization of the filter cavity lock.
First we have experienced some trouble with the labview ADC of the local control signals for the input mirros. The error signals for the pitch and the yaw looked the same. After some attempts the problem was solved simply by switching off and on the CPU unit. We have already observed this problem in the past but I could not understand what triggers this kind of behaviour.
After recovering the local control we could align and lock the cavity.
At first we wanted to measure the calibration factor for the error signal. In order to do that we summed a line to the piezo correction (input channel " RAMP") above the UGF.
We monitored the amplitude of the line on the monitor channel "PZT mon" finding a value of V_RMS = 17.84e-6 V.
The correspondent amplitude in Hz is
S_Hz = V_RMS (V) * 100 * sqrt(2) * 2e6 Hz/V = 5045 Hz (1)
The factor 100 is the attenuation of the "PZT mon" channel
The factor sqrt(2) is to pass from V_rms to the line amplitude
The factor 2e6 Hz/V is the gain of the piezo after the SHG
The correspondent amplitude observed in the error signal is
Err_V = K(V/Hz) * S_Hz/sqrt(1+ (f/f_0)^2) = sqrt(2)* 0.465 V (2)
% (2020/09/04) Note that this formula should be Err_V = K(V/Hz) * S_Hz/sqrt(1+ (f/f_0)^2)
Where we have taken into account the effect of the cavity pole assumed to be at f_0 = 1.45 kHz.
Comparing the amplitude of the two lines (see picture 1), inverting Eq. 2. We could find the calibration factor K (V/Hz) = 2.5 e-3
We have used it to calibrate the error signal into Hz and we obtain a lock accuracy of about 120 Hz, (see picure 2) which is similar to what we obtained in July.
We have also performed a noise injection to measure the open loop TF finding a UGF at about 10 kHz with a phase margin of 50 deg. (See picture 3)
We have tried to change the gain but we observed that it was already optimized to have the error signal as small as possible.
In picure 4 the transmission of the filter cavity is plotted.
Since we found something wrong last week, we tried to carefully measure the beam parameters again after the first Faraday Isolator. As Yuefan did, we also set the origin as the BS on the bench.
For x direction, the beam waist is 118.2 um at the position z=-0.10 m. For y, the beam waist is 119.8 um at z=-0.093 m. See attached figure 1 for the fitting result.
Then we used this result to simulate the propagation of the beam using the software "Jammt". See attached figure 2 for this simulation.
We found that our beam parameter was very sensitive to the last lens position. We have fine tuned its position in order to have a reasonable beam dimension. By using a steering mirror on the beam path we could propagate it for several meters in the central area and check that the beam size should be reasonable at the input of the in vacuum faraday.
After that we have keep trying to recover the alignment of the infrared without success. We have collected as many cameras and screens as possible as use them to look insides the input vacuum chamber simultaneously.
Despite many tries the situation is not different from that of the last time:
1) We can recover the references for the IR on the plastic film but for one of them the beam as a quite irregular shape.
2) We could see the beam on the mount of the 2 inch mirrors and we tried to center it.
3) We could see the beam hit the leg of the PR telescope mirror and tried to center it on the PR.
After that we cannot see the beam on the BS or on the input mirror but just some strange quite dim shape on the first target.
As already observed in entry 631, when moving the PR or BS mirror with the local controls we can move accordingly the shape on the target. This is different from what we expect if the beam was hitting the pipe, since normally in this condition we observe a change in the intensity and in the shape of the scattered light.
Conclusions: We are probably not understanding what is happening inside the chambers and we need some new ideas to go on.
For the definition of the incident angle, please refer to figure 3.
For the spec of the grating, please refer to the attached pdf file.
I am not really sure about what I see on the graph.
I need your opinions.
Because yesterday's measurement doesn't match previous result. I decided to do the simulation for the infrared. However, I found there is a lens without label on it. So Eleonora suggested me to consult yuefan about this. And yuefan said it is used for the test of mode cleaner. So we decide to remove it. After removing it, we want to check the beam waist again.
See attached figure 1, we found the beam waist was very far from where we did the measurement.(The measurement position is shown on the attached figure 2) And the beam waist is about 1mm. The beam waist fit the result of OptoCad. However the beam waist position doesn't fit the result of OptoCad. As you can see the attached figure 3, the highlight part show that the beam waist is around 8 meters ahead of R1. This R1 is the dichroic mirror which is used to combine green and infrared. But the measurement of beam waist tells us the distance between them is around 6 meters.
According to attached figure1, the beam size at Faraday Isolator should be around 2mm in radius. This measurement is totally different from the requirement of e-log 441. E-log 441 gave us the beam size around Faraday Isolator should be 1.368mm. No matter it is diameter or radius, it doesn't agree what we measured today.
Besides, I just found there is e-log 442 which tells us the setting of infrared telescope. We can see from there, the design is using leses f500, f300 and f350. However I found the actual lenses we are using is f500, f150 and f175. See attached last three pictures to check this.
Today I measured it again. Since the last measurement seems to be limited in a small region, I tried to expend the region this time. The result is shown here.
And now, we have four lenses on the infrared path. They are f500, f150, f175 and one without lable. Is this the same with previous situation?
we found the pd DET10A for the 633nm probe didn't give any output. Replacing the power supply with a 12V battery solved the problem. We checked the power supply with a multimeter and it looks fine (9V of output). We figured out that it's the adaptor bad contact. We continue working using the battery.
We made a scan of the surf ref sample and maximized the AC signal adjusting the pump alignment.
Current experimental parameters:
Laser current 1.3A Power meter value: 31mW (without sample)
DC signal: 3.26 (with sample) (3.4V at the max of the scan)
AC signal max: 0.2V
After doing that we have put to zero the local control error signals and we have closed all the local controls loops. Then, we could fine tune the alignment with the local controls and lock the cavity stably.
We took pictures of each local control panel while the cavity was locked (See figure from 1 to 4: END, INPUT, BS, PR). The lock can last for more than 30 minutes (we had to unlock on purpose before going away). See figure 5 for the transmitted beam during the lock. We took photos of the oscilloscope showing the transmitted power just after achieving the lock (pic 6) and after 30 min of locking (pic 7). In the best alignment condition it is about 1 V.
So I did experiment today. Fortunately, I found a PBS without costing a lot of time. Although yuefan told me that the most consuming part of job in TAMA is finding something.
Then I improvised the testing set-up. I found most of the light is reflected by the PBS. However, I didn't know the ratio between s and p polarization should be around 1000, which I learnt from Matteo latter on. But this can explain why the energy doesn't conserve if I put half-wave plate in-between. You can refer to attached picture 1. I will take the picture of weak light and measure it next time. As Matteo suggested, the weak signal will have less fluctuation which is important for measurement.
The power evolution is shown as attached picture 2. The attached picture 3 is about the diffraction angle.
I will do the measurement of green soon and compare it with this one. Before that we will refine the alignment of some wave plates.
I use the holes on the bench to estimate the incident angle. The test are divided by two sorts. For the first one, the incident beam is aligned with the arrow on the grating. For the second one, it is opposite.
I also test it by using a wave plate. It is used to see if polarization can affect the diffraction and if so, how it can affect.
In the entry 613, I made a miscalculation. This entry is a correct one.
I re-calculated optimal lens pairs for a telescope for the mode cleaner (green) with accurate dimensions (see here for the drawing). Figures attached show lens pairs that seem acceptable and the parameters I used.
In the case of the first figure, the mode matching factors are 99.718% and 99.601% for vertical and horizontal axes, respectively.
In the case of the second figure, the mode matching factors are 100% and 99.727% for vertical and horizontal axes, respectively.
The lens with 75.6 mm focal length that we already have is one without any surface coating thus low transmissivity. So I think the first case is preferable.
Figures were missed in the entry. They are shown here.
Open-loop transfer function
On 12/21, we measured open-loop transfer function of SHG control, because a new servo was designed and installed by Matteo Leonardi. For the previous situation, please refer to an entry 585.Fig.1 shows open loop transfer function of the SHG control. We have three switches on the new servo, each of them is corresponding to a different integrator. Measurements were conducted for three combinations of these integrators. Noise level inserted were different depending on the conditions in order to keep the cavity being locked. PZT resonance point is located aroun 25 kHz. Unity gain frequency is shown on each graph by a black line (770 Hz, 1380 Hz, and 1610 Hz). You can see a high phase margin (approx. 90 degree) when the mid integrator ON.
We also measured power spectra of an error signal on the loop and DC signal from the photo detector which is monitoring transimitted light from the SHG (Fig.2). You can see a spike around 600 Hz. I am not sure what it means.
All spectra were taken by Agilent 36540A.
Output power stability
Fig.3 shows a long term (1000 second) stability of the SHG output. Stability of the SHG is much better than before. In this moment, we do not find a large fluctuation in the SHG output (532 nm) as we did so far.
Figures were missed in the entry. They are shown here.
A comment on the local controls large output: the voltage range at the output of the DAC is +/- 10 Volts. So If the correction signal is out of this range (as in the case of picture 1) the output is saturating and the control loop is likely not to be working properly (as can be seen from the large RMS in picture1).
More in general, looking at the error singnal, it seems to me that we are still affected by the "spikes issue". It could be useful to compare the open loop rms with the reference values in the absence of spikes.
We know for PDH control, the local oscillator phase and signal phase must match. But the cable length can affect this phase. So after changing cables, we need to set the phase of signal generator again.
Firstly, we need to close the loop for local control to make the light beam interferes well inside the cavity. However, I found the local control goes oscillation even with the previous value. The previous value means the value we used for locking yesterday. I should mention this abnormal case happened the day before yesterday.
After a very long time of adjusting, I found the the local can be fairly stable while the output of local is large(Fig 1). I adjusted it a little and make the output much smaller(but it is still around 40). Then I could find the beam through the end room camera.
Secondly, I took the photo of beam on the first iris(Fig 2,3) and the second iris(Fig 4).
Thirdly, I went to set the PDH phase. I set different value to see how the error signal looks like. I took three pictures(Fig 5,6,7) for this processes. We can deduce from them the worst case should between 80 and 90 degrees. So I set the phase as 175 degrees.
Finally, I found the transmitted signal decreases from 1V to 0.6V(Fig 8). And I found this is not caused by the SHG. Now the SHG locking is pretty pretty good! This is caused by the bad local control. The position of mirror drifts away fast. However, this maybe also caused by the large output of local control. Besides, I and Matteo measured the TF of BS pitch. Matteo said it is not a good control system according to this TF. So the local control definitely should be improved.
A comment on the local controls large output: the voltage range at the output of the DAC is +/- 10 Volts. So If the correction signal is out of this range (as in the case of picture 1) the output is saturating and the control loop is likely not to be working properly (as can be seen from the large RMS in picture1).
More in general, looking at the error singnal, it seems to me that we are still affected by the "spikes issue". It could be useful to compare the open loop rms with the reference values in the absence of spikes.