Last Wednesday, we worked on the improvement of the error signal coming from the IMC end mirror quadrant photodiode which looked spoiled by scattered light noise. Indeed, we already identified at the installation some extra light coming from MC tower which was focused in a point not far away from the main beam focus and it was somewhat difficult to get rid of it. We have made several checks (using a video camera) without understanding well the origin of this light. In order to get rid of part of this spurious light, we have installed a diaphragm of 50mm clear aperture right at the output of the tower window. We accurately positioned it by looking at the beam on the west wall of MC building. In the end, we installed a new quadrant photodiode on the bench the replace the old one which was a temporary one. The improvement on the error signals is significant even if it looks that there is still some residual scattered light noise. We are planning some noise injection to try to understand the origin of this noise and maybe on a longer term to install some baffles inside MC tower to kill the scattered light coming from the tower.
The IMC AA loops have been left open for the night since we couldn't complete the activity planned around IMC end mirror quadrant
which consisted of studying the spurious light reaching the quadrant and replace the quadrant by a new one. The work will be completed tomorrow.
IB and MC are under local control and BPC loop is closed.
Now that the new dihedron has been installed we have repeated the absorption measurements that were carried out in logbook entry 31732 with the "Virgo " dihedron. As before, the meausrement consisted of changing the laser power at the IMC input and we measured the frequency shift of the HG10 and HG01 mode. The results are reported in the following tables and the frequency shift versus input power is plotted in the figure.
In the logbook entry 31732 we saw that for the HG10 mode the frequency shift induced by absorbed power is almost the same for both the end mirror (~32.5Hz/mW)Â and the dihedron mirrors (~35Hz/mW for each dihedron mirror). As we cannot distinguish effects caused by individual mirrors we will consider that a shift of 34Hz corresponds to a total absorption of 1mW in the three mirrors. The measured frequency shift change from 3.43W to 7.38W input was 177.5Hz. To account for this 177.5Hz, total absorption needed is 177.5/34 = 5.22mW.
The cavity power change is (7.38-3.43)*(1160/pi)=1458 W, so the total absorption is 5.22e-3 / 1458 = 3.6ppm. This compares to 3.1*3=9.3ppm that was found using the "Virgo+" dihedron. We may therefore assume that the absorption of each dihedron mirror has reduced by (9.3-3.6)/2=2.9ppm.
A FFT simulation has been used to calculate the shift in frequency of the HG10 and HG01 modes due to the deformation of the IMC mirror surfaces caused by coating absorption. The simulation includes the DIHEDRON geometry, the baffle with two 40mm holes in front of the DIHEDRON and phase maps (provided by LMA) of all three mirrors. The thermal distortion of the mirror surfaces was calculated using the Hello-Vinet formula, which was confirmed to be accurate enough for all three mirrors by comparing with COMSOL calculations.
The results of this simulation are shown in the figure. We see that the frequency shift for the end mirror heating is the same for both HG10 and HG01. The frequency shift for the DIHEDRON is almost the same as that of the end mirror for the HG10 mode (horizontal plane). However, the two frequency shifts are different for the HG01 mode (vertical plane). It is not easy to give a simple explanation. One key issue is that the thermal bump of the DIHEDRON is almost axi-symmetric, and the size of this axi-symmetric bump looks like a horizontally wider oval seen by the beam.
It took quite some time to find the right way to measure the position of the HG01 and HG10 modes. The setup used for the measurement is depicted in the first picture. The IMC cavity is locked as usual phase modulating the laser beam at 22.304MHz. The new feature used consisted in using the second EOM to scan the cavity while being locked using a second pair of sidebands with much smaller modulation depth (a sweep is generated using a HP network analyzer). Looking at the RF IMC error signal (in reflection) looking in the Free spectral range around the modulation frequency (above 21 FSR) you can see that when you tilt the input beam horizontally by adding an offset on the Beam Pointing control system (opening the AA loop of the IMC cavity), the 10 mode amplitude increases. An example of the HG10 amplitude displayed on the spectrum analyzer is shown in the second picture. The experiment we did consisted in changing the laser power at the IMC input (from 4.9 to 7.2W) and we measured the frequency shift of the HG10 mode. The results are reported in the following table.
NB: the FSR has been measured (see entry #31678 (https://logbook.virgo-gw.eu/virgo/?r=31678)) to be 1045125.3Hz.
A FFT-based simulation was used to estimate the absorption rate of mirrors based on this frequency shift. The simulation includes the DIHEDRON geometry, the baffle with two 40mm holes in front of the DIHEDRON and phase maps (provided by LMA) of all three mirrors. The thermal distortion of the optic surface was calculated using the Hello-Vinet formula, which was confirmed to be accurate enough for all three mirrors by comparing with COMSOL calculations.
The HG10 frequency shift was calculated for different cases:
1) By heating the end mirror by 1mW absorption, the frequency shift is -32.5Hz.
2) By heating two DIHEDRON mirror by 1mW each, frequency shift is -70Hz (~2x of the end mirror effect)
3) By heating all 3 mirrors by 1mW each, frequency shift is -102Hz (essentially the sum of case 2) and case 3))
These results show that the thermal distortion changes the frequency by ~34Hz/1mW per mirror. The current experiment cannot distinguish effects caused by individual mirrors and 102Hz / 1mW-per-mirror is used assuming the absorption rates are the same for all mirrors. The measured frequency shift change from 4.87W to 7.22W input was 270Hz. To account for this 270Hz, total absorption needed is 270/102 = 2.65mW-per-mirror.
In order to estimate the power change in the cavity, we have used the last numbers which were measured extracted form the cavity pole and the transmission of Input and output dihedron mirrors.
T=81.3%+/-2.3% (Throughput)
Matching=0.945 (IMC matching)
PoutIMC=7.6W*Matching*T=5.84W+/-0.17W -> this corresponds to about 7.44V on IMC_TRA photodiode.
This means that the conversion factor for the photodiode is about 0.785W/V. Thus, during the experiment the power was varied between 4.87 and 7.22W-> DP=2.35W.
The cavity power change is 2.35W x (1160/pi) =867 W, so the absorption is 2.65e-3 / 867 = 3.1ppm / mirror. The throughput and the mode contents are calculated for the case that each mirror absorbs 100mW, for 90W input power, and the thermal effects were found to be negligible. With 150mW absorption, for 135W input, the throughput of the TEM00 mode drops by 3%. With the simulation setup described above, the round trip loss is calculated to be 80ppm and the observed 400-600ppm loss cannot not be explained. A small misplacement (several mm) of the baffle in front of the dihedron increases the round trip loss only by several 10ppm and doesn’t seem the be to cause of the extra losses since we measured similar round trip losses without this baffle (see logentry 31636 # https://logbook.virgo-gw.eu/virgo/?r=31636).
The simulated value of the HG10 frequency does not match with the measurement. The RoC of the end mirror based on the frequency measurement is 183.7m with 4.9W input power, but the RoC using the frequency shift by the simulation is 185.4m at the cold state, with 185.1m RoC being used in the simulation. This discrepancy has to be further understood. Similar measurements will be carried out on the IMC cavity with the new dihedron in the next days.