According to the basic theory of the PCI method, the heated area of the sample makes a phase shift in the probe beam; this perturbation is small and can be treated as a gaussian beam which interferes with the main beam;  the maximum of the interference is detected when the PD is at the Rayleigh length of the gaussian perturbation, which can be calculated using the waist of the perturbation, which is the pump beam size.

The approximation of the perturbation to a gaussian beam is valid at first order, but for a fine tuning of the detector position,  it might be not a good approximation. I consider this because, when I correct the position for the thick sample (as in elog entry 291), I notice that the calibration value is not the same (as expected from the simulation). A possible explanation might be that when I put the thick sapphire sample and correct the Image unit position, the detector is not in the interference maximum anymore. So, I  maximize the signal as a function of the detector position experimentally, by moving the Image Unit with the micrometer screw. I do it for the reference sample alone, and for the reference sample with the sapphire sample behind it. Then I compare the two maxima positions in order to find the best position correction for thick samples.

First two plots show several scans of the reference sample for each position of the Image Unit, with and without tama sapphire sample. In last  plot, I took the middle value of each scan and plot it as a function of the Image unit position. 35mm is the closest position of the IU to the sample, 0mm is the furthest IU position. To move it further it's necessary to unmount the IU micrometric translation stage. The theoretical distance between the maximums is 26mm. and the maximums should have the same value.

The plot shows  maximums position accuracy of about 5mm, but in the case of reference + tama sapphire, it's not clear wether the maximum is below the position 0mm on not. The problem is the maximum value, it should be the same but for the reference alone the value is 0.1 and in the other case is 0.04. More than a factor of 2

I repeated the measurement of entry 289, but this time, I put the tama-size sapphire sample in a position such that the probe beam is crossing it but the pump beam is not, so I avoid any back reflection of the pump. I also correct the position of the Image Unit according to the formula I wrote in entry 290.

I moved the base micrometer by the distance L x (n-1)/n - 1mm away from the sample. L is the path inside the sapphire sample, which is the thickness 60mm divided by cos(6°), 6° is the probe incidence angle, n is the sapphire refractive index 1.76. Therefore the displacement is 25.05mm

The attached plot shows the comparison of the two scans.

The result is not as good as in the simulation. The DC is different because there is an additional reflection when I put the second sample, but the AC/DC should be equal in the two cases.

I think I have to figure out why. Maybe the positioning should be slightly different, so I will try to find he optimal position.

I got a formula to correct the positioning of the detection unit.

 SampleThickness*(n-1)/n - 1mm

In the case of 60 mm-long Sapphire the shift is 60*0.76/1.76 - 1 = 24.9 mm

I applied to the last simulation and it looks working.

I tried to reproduce experimentally the situation of last simulations (Elog entry 288), but the data are a bit confusing. I think I'm missing something.

Here is what I did:

I made a scan of the bulk reference sample (blue data in the plot), and then I placed the tama-size sapphire sample in front of it, at about 15mm, and repeated the scan (black data in the plot).

Pictures of the setup: 2,3

With the sapphire sample, I noticed that:

 - the signal is very low

 - the phase signal has a strange shape.

This phase shape makes me think that the sapphire sample might reflect the transmitted pump beam back and then heat the reference sample again. This would change a lot the signal shape. So I placed the sapphire sample further, at about 45mm, and a bit tilted (about 3°) so the reflected beam would not go back on the measured point. Then I repeated the scan (red data in the plot). Picture 4. Then I tilted more the sample (about 6°) and repeated the scan (green data). I couldn't tilt the sample more because, otherwise, the probe beam would go out of the prism mirror.

Every time I moved the sapphire sample, I had to tune the focusing lens to center the probe beam on the detector.

I feel confused by the fact that changing the position of the sapphire sample make so different signals.

In order to calculate the calibration in the case of thick samples, I simulated the scan of the bulk reference silica sample, then I simulated the same thing but adding a 6cm-thick sapphire sample on the probe path.

I already calculated the probe beam size on the detector for different sample thicknesses (Elog entry 263)

In the  first plot, there is the scan of the bulk reference sample (red line), and also the same scan but with a 60mm thick sapphire after the sample (blue line). Adding 60mm of sapphire after the sample changes the optical path of the probe and makes a different signal.

The calibration value is taken at z=2mm, and there is a factor of 5 of difference between the two cases.

The second plot is a 2x2 matrix of plots and it shows the probe beam profile at the detector, when the sample is at z=2mm. First column of plots is the beam profile. Second column is the interference pattern, from which the AC signal is calculated. The first row is the case with only the bulk reference sample and the second row is the case with the  bulk reference plus the tama-sized sapphire sample after it. The white rectangle is the profile of the photodetector. The unit of axis is m, the photodetector is 1mm large.

I think it is necessary to adjust the reimaging of the detection unit. I will try to get better signal in the simulations by changing some distances among the components and also changing the focal lengths of the lens and of the sphere. 

I will also reproduce the same experimental configuration of this simulation, putting the tama-size sapphire sample after the reference sample, and making a real scan. This will also be a test to see how reliable is the simulation.

In the following table, there is a summary of last measurements

"Depth" is the position of the incident sample surface with respect to the pump-probe cross point. According to the refraction effect, the cross point position inside the sample is double, so, given the sample thickness 60mm, the depth 30mm refers to the measurement at the opposite surface of the sample.

I attach the plots of the 6 sets of measurements.

Today I tried to do some measurements on a plane Beam splitter with the JASMINE setup.

Unfortunately, neither the laser nor the monitor were working... On the laser's power supply also no error signal has been shown. Just no signal at all. 

I wonder whether this is due to the humidity...

In order to test the configuration of the optical devices that will be used for the BS-OpLev in KAGRA, I simulated the basic setup of it on an optical table in the ATC.
For the position measurements, I have used a PSD (Position Sensing Detector) from Thorlabs (PDP90A).
The calibration of the PSD could be done in Y-direction only as the X-direction is not accessible due to the setup (for this I would need a 3-axis mount; the one used in the test was just a 2-axis mount). The resulting function Vy/Vsum is linear along the y-axis with a mean gradient of ca. 193 1/m (for comparison, the number that Eleonora has measured is 184 1/m for a PSD of the same type in TAMA).
A graph of the measured data along with a fit is shown in the attachment. Also shown are photos of the setup and a sketch of it.

I'm scanning the sample along the beam direction. I placed a translation stage with a micrometric screw below the mirror mount and set the sample height so that the pump goes at the centre of the sample.

Pictures 1,2,3 show how it was before.

Pictures 4,5 show the current setup.

I'm doing one measurement for each 5 mm of sample depth.

I did the same measurement for the 4th time but this time I covered better the pump path so that the scattered light is less.

The absorption value now is 4ppm/cm with a precision of 0.8 ppm/cm (after filtering).

This means that the pump stray light (1064nm) from outside the box gives a contribution of at least 2ppm/cm on the measure.

I couldn't cover the sample, but also the scattered light from the sample could give a significant contribution. So I would like to put a filter in front of the photodetector, to make only the probe light (633 nm) pass.

I ran another measurement with same experimental conditions as the ones in (entry 279), and made the same analysis of (entry 280) and (entry 281)

2016-07-25. Tama-size , 1h, rate: 100ms.

after filtering

I analyzed data of the two last measurements (entry 279), I made groups of 600 samples (1 minute) and I made the histogram for each group and the gaussian fit of the histograms.

I plot the fitted parameters of X/DC signal and Y/DC signal as a function of time (for each minute).

In blue the 0W pump data; in red the 9W pump data.

The thick line is the mean, the dashed lines are the mean ± sigma.

I did some analysis on the last absorption measurements (entry 279).

X, Y: output signal from Lock-in Amp.

I fitted the histograms of X and Y signals with gaussian. For raw data and for filtered data.

I calculated the precision of the measurement as std(sqrt( (X-Xo)^2 + (Y-Yo)^2 )) and used the calibration to get the ppm/cm

2016-07-20. Tama-size , 1h, rate: 100ms.

2016-07-21. Tama-size , 1h, rate: 100ms

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

After average filtering, filter order: 600 (1min)

2016-07-20.

2016-07-21.

I removed again the translation stage and placed the tama-size mirror mount, tightly fixed to the optical board.

Yesterday I took 1h of data (rate 100ms) with the pump OFF and  1h of data with the pump ON (9W).

Today I repeated the same measurement: 1h at 0W and 1h at 9W.

I plot the raw data. Blue is the 0W data and red is the 9W data.

I will filter and calculate the ppm/cm.

I have calibrated the PSD used for the optical lever of the filter cavity mirrors. PSD model : PDP90A thorlab (see file attached).

I computed the calibration for 4 different values of the power (V_SUM should not exceed 4 V)

Mean = 0.00543 m  std = 0.00008 m

Normalised calibration seems reasonably independent of the power.

In order to recover the appropiate calibration ( m/V) to convert a voltage signal into a displacement of the beam on the PSD, this value shoud be dived by the measured V_SUM.

After closing the control loops of pitch and yaw of the telescope mirror in the PR tank, I tried to calibrated the signal in order to have an estimation of the angular motion of the mirror.

The displacement of the beam on the PSD given to a rotation in yaw on of an angle delta is given by

x  = 2* arm* delta

Where arm is the distance between the mirror and the PSD.  

A dispacement of the beam on the PSD of an amount x corresponds to a PSD voltage output of  x/cal.

So  delta = Vout*cal /(2*arm)

I measured the calibration factor of the PSD for 4 different powers in order to check if the normalized calibration (that is the calibration divided by the PSD voltage sum)  was constant.

I found 

mean (n_cal) = 0.0071      std (n_cal) = 0.0002     In order to recover the appropriate calibration (m/V) this value should be dived by the V_sum measured each time.

Assuming arm = 1.20 m +/- 0.15 m  (the precison on this measurement can be increased by measuring the optical lever arm next time we open PR tank)

and having measured V_sum = 13.3 V

we have

Cal_tot =  n_cal/(2*arm*V_sum) =  (2.2 +/- 0.3 ) e-4 [1/V]  (percentage error 13% to be improved by better measuring the arm)

The comparison between the calibrated spectra with open and closed loops for pitch and yaw is shown in figures 1 and 2.

Some remarks:

1)In both cases  the RMS seems to be dominate by the displacement in the region between 3-10 Hz.

2) Since the optical lever makes use of two steering mirrors directly fixed on the stack (see entry 262), this could not be a real motion of the mirror but a motion of the stack

3) This shoud be understood in order to improove the filter shape (Shoud we gain in that region or not?)

NB. For the pitch calibration we need to take into account an additional factor, equal to the the cosine of the incidence angle of the beam on the mirror (see 3rd attache picture)

y  = 2* arm* delta* cos(alpha)

THIS FACTOR  HAS NOT BEEN TAKEN INTO ACCOUNT IN THE PITCH PLOT WHICH SHOULD HAVE BEEN MULTIPLIED BY A FACTOR 1.412 (since alpha is 45°)

I ckecked the dimensions of BS intermediate mass.

I found that clamp parts are common for NM, EM ans BS itermediate masses.

Therefore, I can conclude that a hanging jig for NM and EM mirror can use for BS.

Since the fans of the booth have been off for a couple of months. I had to clean everything from the dust. I started from the top shelves, wiping one by one all the objects. I moved the boxes made of paperboard out of the clean booth because paperboard  is known to produce dust. I cleaned the optical table and all the objects on it. Wiping with a wet tissue was not enough because the tissue releases fibers and dust. So I used the strong green lamp to watch the dust particles, the spray air to blow on the surfaces to make the dust fly and the vacuum cleaner to blow it up from the air. The vacuum cleaner was outside, I only brought the pipe inside.  After that, I measured again the particles number.

This afternoon I measured the spectra and the transfer functions of the mirror installed on in the PR tank (to be used as a part of the injection telescope of the filter cavity).

The mirror motion in pich and yaw is sensed by means of a optical lever. The spectrum of the motion in the two degree of freedom is shown in fig1 and 2 of the attached file.

In order to measure the transfer functions, I injected white noise in pitch and yaw (with an amplitude of 3 V ).  To improve the diagonalization of the sensing, I changed the the sensing matrix appying a rotation of 0.04 rad. I have also slightly changed the driving matrix, to reduce the excitation of the yaw resonance when injecting noise on pitch. The comparison between the TFs before and after these changes are shown in fig 3-4 and 5-6 for yaw and pitch respectively.

When we install the BS suspension,

(1) one magnet came off, and

(2) the lower suspsnsion wires were broken.