Takano, Tanaka, Miyakawa

Abstract

We recovered the RFAM monitor, but for monitoring it in DGS some preparation of the real time model is necessary. We tried optimizing the polarization of the laser injected to the first EOM with 1 W of PMC output, and now the polarization is almost p-pol. However, after increasing the power from 1 W to 14 W, RFAM seemed to get worse. Further tuning of the polarization is necessay.

Detail

To tackle many issues related to RFAM, we started the investigation of RFAM.

Preparation

First, we lower the laser output down to 1 W at PMC output, and checked the RFAM monitor already installed in PSL. We confirmed that the beam path going to the RFAM monitor is as designed (Figure1). However, we found that:

• The DC power supply for the RFAM mionitor was off, and even its power cable was removed.
  • We added a power strip next to already installed one to connect the power cable of the DC power supply because there was no outlet availabe on the old one.
  • A new power cable was conneced to the DC power supply.
  • We confirmed that the RFAM monitor works and the alignment is optimized.
• An I/Q demodulator was installed around the RFAM monitor and seems to demodulate the RF signal at f1, f2, and f3, but its power cable was removed.
  • We installed a new power Dsub cable and connected it to the demodulator.
  • We turned on the switch, but we couldn't able to confirmed whether it works or not, because at that time we were not sure which channels we had to see to check it.

After these preparation works, we took a Moku:Lab (Moku PSL) from a rack in which TTFSS is installed to another one close to the duct and connect the AC output of the RFAM monitor to Moku. Figure2 shows the spectrum up to 89 MHz. There were shar peaks at 13.7 MHz (f_imc), 45 MHz (f2), and 80 MHz (f_aom). When we blocked the light going to the PD, still 45 MHz and 80 MHz peaks (Figure3). When we disconnect DC output of the PD, these two decreased (Figure4). Because we don't need the DC signal so far, we kept it disconnected.

Check of oscillation signal

Next, we checkded the RF signal applied to EOMs (Figure5). Although there is some offset in each modulation signal, it seems not so strange. We think that they are healthy.
During this work we noticed that EOM3 moved when we connected the SMA cables with a torque wrench and went back to the original position by springs of the EOM mount. We saw EOM3 with an IR viewer (actually its was a camera) and there seemed to be no clipping. We also checked the power of IMC REFL, and confirmed that the power was not changed a lot. Therefore, we concluded that the alignment didn't change during this work.

Optimization of the polarization

Next, we started to adjust the polarizaiton of the light injected to the EOMs. First, we tried tuning the polariaztion as pure s-pol as we could. We put a QWP and HWP with this order after PMC, and put a cube PBS after the HWP. We aligned the angle of the wave plates so that the transmission power of the PBS is minimized. At that time, the angle of the HWP read 270 deg, and the light after HWP should be s-pol.
After that, we removed the PBS and checked the spectrum of RFAM monitor. The spectrum as that time is shown in Figure6. We tried changing the angle of the HWP while monitoring the height of the peaks, and we observed:

• When rotated in negative direction (270 to 260), the height of the peaks increased (Figure7).
• When rotated in positive direction (270 to 315), the height of the peak decreased. The minimum height was achieved at the angle of 318 deg (Figure8).
• While rotating, sometimes the peak at fimc decreased whereas the peak at f2 increased (the mdulaton signals at fimc and f2 were applied to thesame EOM).

Considering the angle of the HWP before this tuning, the rotated angle is about 48 deg, which means that now the polarization of the light is almost pure p-pol. This implies that there are some misunderstandings about the nominal polarization in PSL.

Long time measurement

After the tuning work, we increased the output power to the nominal value (14 W at the output of PMC). While increasing the power we kept watching the space around the wave plates we installed with the IR viewer and seemed no accident. The RFAM spectrum at that time is shown in Figure9. We took the same data 3 min. later (Figure10), and found that the spectrum changes with time. We thinkthis is due to the temperature change induced by human activities or absorption of the high power laser. We did some cabling work to connect Moku to the network, and left PSL room without changing the setting of the wave plates.
From the outside of PSL we aligned PRM and checked the demdulated signal at REFL port in the same way as the previous log. We rotated the HWP of the power controller by 20 deg as shown in Figure11, to set the incident power to IMC as the current nominal value. It is strange because without any polarizing optics the rotation angle should be coincident with that of the HWP before EOMs (48 deg).
Figure12 shows the time series after today's work. We found that:

• Still we can see the offset value in the demodulated signals. They seems to have some correlation with the temperature of PSL (K1:PEM-TEMPERATURE_TABLE_PSL_PT04).
• The output of PSL (K1:LAS-POW_PSL2_OUTPUT) also drifted and have correlation with temperature. 
• The DC power of each RFPD also drifted.

The RFAM spectrum when we started this measurement is shown in Figure13. The signal of RFAM PD got half because before leaving PSL we divided the signal by a power splitter for future usage. This spectrum was taken about 45 min. after taking the data of Figure12. 

Discussion

• In the low power case, the optimal polarization to EOM1 seems p-pol, while for a long time we injected s-pol to EOM1. From this document, the EOMs seem to be designed so that p-pol light is injected. Also, from aLIGO document PMC is designed to resonate in p-pol. However, we adjusted the input polarizaiton to s-pol before PMC. There are apparently confusions about the designed polarizationin PSL.
• Judging from the inconsistence of the rotation angle between the HWP before EOMs and the HWP of the power controller, the polarization state changes between these two. In my opinion, either EOM1 or EOM3 generates large ellipticity and the major axis changes. The change of the major axis also couples to the drift of the output power of PSL.
• Now the polarizaiton of the light injected to EOM1 is not optimized. This may be because the polarization state changed by heat absorption of the wave plates or any other effects after increasing the input power.

Next plan

1. Modify MZI model or create a new model to monitor RFAM continuously with DGS.
2. Test the optimization of the input polarizaiton with low power and measure the long time trend.
3. Check the polarization between EOM1 and EOM2, and after EOM2.
4. Consider the layout of the additional optics (wave plates, polarizers, picomotors, ...) as soon as possible and order these components.
5. We should confirm the nominal setting for the polarizaton of PSL (PMC, EOMs, ...). Need to ask to Nakano-san, Kokeyama-san, and Uehara-san?