NAOJ GW Elog Logbook 3.2


with the parameters reported in elog entry 1239 we did the scan and the map of the black sapphire sample from Shinkosha.
Since the absorption is about 80%, there is a heavy depletion, so the scan is not symmetric.
The calibrated map looks different from the screenshots. It is because the screenshot shows the AC signal only, while the calibrated map is proportional to AC/DC, where DC is not constant because the probe is absorbed not uniformly.

[Matteo, 2*Eleonora]
We tried to check the effect of CC loop on the squeezing.
While bringing up the whole bench we had some trouble acquiring and keeping the lock of the CC PLL.
By modulating the green phase (10 Hz, 1.5 Vpp) we obtained a nice error signal for CC in reflection from OPO. Since the amplitude of such signal is maximum when the parametric gain is maximum we used it to find the optimal freqency offset for the p-pol PLL lock. It is 170 MHz and the correspondent CC error signal amplitude is ~ 65mV pp.
We centered the error signal around zero adding an offset and stopped the green phase modulation. See pic1.
We closed the loop (acting on the green phase shifter) using two stanford SR560 in series with the following parameters:
cut off | gain | order | |
1- SR560 | 5 Hz | 200 | 1 |
2- SR560 | 30 Hz | 100 | 1 |
With only one SR560 we were not able to make it work.
We checked the CC error signal from the homodye: it shows a strange, not sinusoidal shape (even after checking and optimizing alignment). See pic2. We tried to close the loop on the LO phase shifter with one SR560 but we didn't succeed.
We measured squeezing at zero span at 200 kHz with the CC loop on the green phase closed and scanning the LO phase. Even if we saw the usual McDonald shape modualtion in the shot noise, its value was much higher than the shot noise level reference without squeezing. This is probably due to the CC beam reaching the homodyne. We will investigate the possibility to reduce its power.

Scanned the surf ref sample with the HeNe probe - Fig 1
R=AC/DC/P/Abs= 0.54/4.7/0.034/0.2 = 16.9W-1
Scanned the surf ref sample with the 1310nm probe - Fig 2
R=AC/DC/P/Abs= 0.086/1.98/0.034/0.2 = 6.4W-1
When changing the samples we wanted to have about the same DC signal on the PD so we changed the current of the 1310nm laser source.
1310nm parameters:
laser source current 0.7A -> PD DC: 2V (ref sample)
laser source current 1.97A -> PD DC: 2V (cryst coating sample)
Scanned the crystalline coating sample with the 1310nm probe - Fig 3
Abs = AC/DC/P/R = 0.0018/1.9/10/6.4 = 14.8ppm
Map 4mm diameter - Fig 4
Calibrated Map and histogram - Fig 5

Incident power 36mW
Transmitted power 6mW
x=305.432
y=118.823
zmax=65.55
x=305.432
y=111.823
zmax=65.55
x=310.432
y=118.823
zmax=65.55
MAP
range x: 310.432 - 298.432
range y: 109.823 -124.823
More precise measurement of the reflection/absorption characteristics of the sample. Chopper OFF (cw beam)
Sample position X = 304.432 Y = 117.323 Z = 65.55
Incident: 80+-1 mW
Transmitted: 7.45 +/- 0.05 mW
Refl: 6.3+/-0.3

The incident power on the sample was always P = 10 W.

[Marco Bazzan, Manuel Marchiò, Matteo Leonardi]
This is an update of the measurement campaign on Shinkosha sample S#4 where several time consuming maps were recorded.
Calibration as in entry 1221
We started a XZ map but it looked a little blurred (Figure 1), so we decided to launch a new one with a smaller step along X (Figure2). In that case the detail is better. We then tried a XY map with a small step, resulting in a series of striations with a period of about 50 microns (Figure3).
A final XY map was taken on a larger area (Figure 4).
The incident power on the sample was always P = 10 W.

[Aritomi, Eleonora, Matteo]
First we checked if the alignment of AMC changes or not from yesterday. For LO, peak is 8.16V and mismatch is 31.2mV+4mV. Alignment got worse from yesterday a bit and it was horizontal misalignment (31.2mV). The mode matching was 99.6%. After alignment, horizontal misalignment became from 31.2mV to 2.4mV. The mode matching is 99.9%.
For BAB, first we maximized OPO transmission of BAB by changing p pol PLL locking frequency. P pol PLL frequency where OPO transmission of BAB is maxmized is 288MHz without green.
Then we injected green and maximized parametric amplification by changing p pol PLL locking frequency. P pol PLL frequency where parametric amplification is maxmized is 174 MHz with 48mW green.

gain 2, 20 dB attenuation, 30 Hz lowpass

[Aritomi, Eleonora, Matteo]
First we checked alignment of LO inside AMC. The alignment was very very bad and we couldn't find any resonance. So we decided to remove a front mirror of AMC and aligned from the scratch. While we were doing alignment, we found that a mirror just in front of AMC was loose. That's why we sometimes lost alignment of AMC suddenly. After we fixed it and aligned AMC, we recovered the alignment of LO inside AMC.
For LO, peak is 8.16V and mismatching is 5.6mV+4.8mV+3.2mV. The mode matching is 99.8%.
For BAB, we moved a lens in s pol OPO trans path to improve mode matching. The lens position was 89.5 mm before and now it's 99 mm. Peak is 232mV and mismatch is 9.2mV. The mode matching is 96.2%.
We'll check if the alignment keeps fine or not tomorrow.

[Miyakawa-san, Matteo, Eleonora]
Yesterday, in the afternoon, Miyakawa-san came to NAOJ to finalize the installation of KAGRA DGS.
1) He set up the router to make possible to access the computer from outside. Actually, we couldn't select a fix IP address for the moment so in case the assigned (DHCP) address changes we shoud use the new one for the remote access. The current one is 133.40.117.62.
2) We powered the DAC, AI and AA modules with +/-18 V. The total current needed for the 3 moduels is about 1.2 A so we had to use two power supplies.
4) We connected the AI and AA modules to DAC and ADC with Dsub9 cables I borrow in Kamioka last week.
3) We created a test simulink model and started to do basic tests on ADC and DAC channes. No major problems have been found so far.
Tests will continue in the next days.

[Aritomi, Miyakawa, Oshino (remotely)]
Currently IP address of a computer and a router is obtained by DHCP, so first we have to fix an IP address. To change the network setting of a computer, we edited /etc/network/interfaces, but we can't fix an IP address so far. Miyakawa-san will take over this Sunday.

Participant: Aritomi, Matteo, and Yuhang
We put two ND filters between the last steering mirror before OPO and the last lens before OPO. They are OD0.1 and OD 0.4. So it gives 10^-0.5 = 0.316 factor to both p-pol and coherent control beam.
The reason is to reduce the noise of coherent control OPO transmission coupled into the homodyne detector.

Participant: Marco and Yuhang
After lock coherent control PLL with a 7MHz frequency offset with the main laser, we tried to demodulate the OPO reflection signal with 14MHz. And the OPO transmission is going to the homodyne detector. Then there is demodulation of 7MHz(7MHz local oscillator was connected).
After the demodulation of each signal, we make them go through the filter separately. And then goes to the green phase shifter and the infrared phase shifter separately. After close these two loops, we can basically lock them.
We scanned the green and infrared phase. The error signal for the green phase is around 70mV p-p while 700mV p-p for IR(green power now is 50mW). This may be the reason why we can lock the IR phase much better.
I measured the optical-mechanical transfer function of green phase part. It shows wired behavior. However, we checked together and the measurement strategy looks reasonable. The result is shown in the attached picture and should be further investigated.
Locking condition, for now | |
green phase | Low pass 3Hz Gain 100 |
IR phase | Low pass 1Hz Gain 50 |

Participant: Marco and Yuhang
Since the locking of coherent control PLL plays a very important role in getting a coherent control error signal. When we want to perform coherent control, we suffered a lot from the bad locking condition.
Then we start to consider why we cannot lock coherent control PLL. At some point, we realize, even we used an attenuator to reduce minor peaks. But the PLL system still tries to bring it to these minor peaks. So we decide to remove the attenuator. Now the situation is that the beat note comes from fiber PD and then has an amplification of 18dB.
In this case, we can lock coherent control PLL much better. The locking scheme(named as PLL_CC20190221) is saved in the default folder of all the PLL settings. And the parameters are listed in the attached figure.

From the datasheet of ADF4002 (page 3), there is REFIN input frequency limit from 20MHz to 300MHz. However, we used to send a 7MHz reference signal into REFIN. So in this case, it seems the old locking scheme of coherent control PLL can be improved.
So we decide to use 21MHz REFIN and divide it by 3. This 3 is the value of R.
Today we tried this new locking scheme. However, it still didn't work.
What we observed was PLL locked on the beat note while it goes away easily(Can we put an integrator?). By turning on slow, it can go back to locking point. But there is always overshot(still high gain?).
However, sometimes, we can lock it successfully. So it seems the shape of the locking filters should be improved. Now it works like a not optimized control loop. Maybe we should measure the open loop transfer function of it?

Since we want a reasonably large coherent control error signal. We exchange 'Demod CC' amplification channel with 'EOM SHG+IRMC' channel.
Reason: SHG EOM channel has an amplification factor of 20dB but we are using a 12dB attenuator to reduce sideband amplitude. This means 8dB of amplification is enough.
While CC DEMOD channel had amplification of 14dB. Actually, the more the better. So we decide to exchange it with SHG EOM channel.
After the exchange, we put 6dB attenuator for SHG EOM.
In practice, we could use the AOM channel which has a 37dB amplification. However, I tested it and the AOM channel seems broken.
In the end, the coherent control error signal becomes around 60mV peak to peak.


In order to have the HeNe with the same size as the pump, we want to add 2 lenses between the last 2 mirrors of the HeNe path before the sample.
Attach a picture with the distances from the sample and the screenshot of the Jammat simulation to find the focal lengths of the 2 lenses.

using the calibration reported in entry 1221 we did 2 rectangular perpendicular maps (xz and yz) of the sample S6