I measured the OLTF of the SHG and IRMC using the old spectrum analyser. I used two methods:

1) In FFT instrument mode, apply white noise to Perturb IN, measure Ch2/Ch1 (EPS1/EPS2) frequency response and coherence using a range of noise injection levels. The noise -> FFT method is fast but less precise when it comes to measuring characteristics about peaks.
2) In swept sine instrument mode, apply swept sine to Perturb IN, measure Ch2/Ch1 frequency response and cross spectrum (swept sine doesn't have coherence option)

In the case of white noise injection, lock at 500 mVpk was a bit temperamental but the spectrum could still be measured. Swept sine worked for an initial injection level of 20 mVpk. For values higher than that, it would unlock when the measurement reached ~ 100 Hz (it starts at the max frequency and goes downward).

Magnitude, phase and coherence for the two cavities are plotted for various excitation levels (fig 1-6), and the best result is collected in a 3x1 image (fig 7, 8)

SHG has a somewhat low unity gain of 600 Hz, nearly 100x lower than previous (50 kHz). Phase margin is about 50 degrees. Maximum coherence is obtained for about 200 mVpk noise injection.

IRMC is very low. It seems a bit strange actually. UGF is around 10 Hz and coherence is quite bad for excitation < 500 mVpk, but it also unlocks really easily when excited at this level.

[Marc, Shalika]

Yesterday we checked again DDS3 outputs and got similar results as last time (3178).

We opened the box to check the +12V/GND provided to each amplifiers. We found that DAC0 and DAC3 were not receiving the expected voltage.

DAC3 +12V cable got easily disconnected and we found that DAC0 +12V pin was extremely tilted..

We removed DAC0 amplifier and soldered a new one. We also resoldered DAC3.

Our guess is that all the +12V/GND cables have a strong strain and twist due to the limited space in the board. So we replace the connection to the board output from a SMA female to female connector to a about 5cm length SMA cable.

All the ouputs of the board are now as expected :

DAC0 8.8dBm

DAC1 9 dBm

DAC2 8.6 dBm

DAC3 8.6 dBm

Maybe the strong twist of DAC0 amplifier +12V pin was the reason for all our issue?

Also the board inside is now quite messy. If we want to tidy it we should prepare some custom length SMA cables.

Marc, Shalika, Michael

We looked a bit at SHG transfer function using the new spectrum analyser (Source -> Perturb IN, Ch1 -> EPS2 Out, Ch2 -> EPS1 Out). Just a quick check using FFT with white noise injection showed that the gain of the loop is somewhat lower than before. The frequency response Ch2/Ch1 in the reference level is flat at 20 dB until 10 kHz, but now we drop off fairly fast at about 100 Hz. Switches are set to NON INV, 1/F3, DIF OFF, SIGN -

I found the SHG wouldn't lock properly even though the servo light was green. First I checked the power compared to reference levels. Input IR was fine (~700 mW) but reflected green was very low, < 1 mW. Next I tried checking the mode spectrum but it didn't show anything at all. After flicking the lock switch a few more times I noticed the green power actually dropped when it locked, so based on that I turned the INV switch to NON INV and now it locks as it should (270 mW green). Basically, we have to do some tweaking of the servo because of changing the EOM

We recently purchased a new Stanford Research SR785 Dual Channel Signal Analyser to be used in the FC cleanroom.

It is a bit complicated. One of the main features is that the measurement is independent of the display. This allows us, for example, to have a really precise measurement but with a really rough display, among other things. It can also save a lot more data into its internal buffer in one go, so we can then transfer to disk without having to redo a measurement.

By default it had a really irritating alert message that sounds like a phone ringing, so I turned it off in "Preferences -> Alarm noise -> Quiet". There is another menu option which says "Alarms -> off" but that controls the display of error messages. Very strange UI in my opinion.

It can write data to USB, however, due to weird stuff, it won't recognise storage devices over 8GB. This seems to be a common issue with making old tech forward compatible with FAT32 USBs. The grey USB drive in the FC cleanroom works fine and I already formatted it.

The spectrum analyser formats the USB to pretend that it is several hundred 1.44MB floppy disks. But unfortunately Windows only recognises the first "disk" in the sequence (labelled 000 in red digits on the front panel of the device next to the USB port), so it still runs out of space. Also at one point the spectrum analyser would refuse to re-format the USB, so I had to Full Format on Windows (slow...) and then re-format on the analyser. Shalika says that apparently you can program it to send data over wifi to the computer, which would be better. The manual talks about cable connections only (RS232 and GPIB) but maybe we can find a wifi attachment.

To do: Figure out a good way to remotely import data from the analyser using programmable commands, and then batch convert into some convenient data format. Personally I want to separate headers and data, and the delimiter doesn't matter to me. MATLAB also seems to not mind this operation. But maybe others also have preferences for dealing with data.

Objective: [In continuation to elog 3177]

1. Measuring ringdown decay time

Details:

1. The new RF switch's output was attached to AOM. The diffraction efficiency remained the same as noted previously, i.e. 57% (as mentioned in 3131)

2. The PD used at the transmitted path is  PDA20CS-EC (see elog 3139 for properties). The PD at the transmitted beam is set with a 10dB attenuator.

3.  The amplitude of TEM00 mode in the transmitted signal was 4V. The demodulation phase was set to 185º. The mode matching was observed to be 4.88/(4.88+1.82)=72%

4. The filter cut-off frequency was 3 Hz and the gain was 100. 

5. The laser temperature was 8400±20. The offset voltage varied from -1 to -0.8. 

6. The cavity was able to lock the TEM00 mode (see Fig 1). The voltage of the RF switch control was set to low to cut off the input to AOM. The possible fitting obtained gives a decay time of 1.45 µs (see Fig 2)

[Marc, Shalika]

We removed the 3 broken amplifiers from DDS3 (ie the ones connected to DAC1,2,3).

We confirmed that the ground and +12V were correctly provided.

We soldered the DAC1 amplifier, brought it to TAMA to test the output and measured the expected 9dBm.

We then soldered the 2 remaining amplifiers.

We then measured :

DAC0 -> 6.5dBm

DAC1 -> 9dBm

DAC2 -> 9.6 dBm

DAC3 -> -23 dBm

Because we did not remove DAC0 amplifier before installing the new amplifiers, we suspected that some bad connections could explain the too low output of DAC3.

We tighten better the connections of all amplifiers and this time we got

DAC0 -> -20 dBm while all other outputs were same...

We tried to untighten a little the connection and this time DAC0 -> -36 dBm (other unchanged)..

We need to investigate the reason for this drop (maybe same as what happened before that `broke` the amplifiers?)

Objective: [In continuation to elog 3145]

1. Testing the new RF switch HSWA4-63DR+

Details:

1. The new RF switch has 4 RF terminals (which can behave as both input and output) and 3 control terminals. The RF common port can behave as output or input depending on how we treat the RF terminals.

2. The RF switch was supplied a voltage (using voltage supply) and control voltage to two control terminals (using DC offset from function generator). The RF common port was treated as input and was connected to the output of the RF driver. The output of switch was collected at RF1. All the other ports were terminated using a 50 Ohm terminator. I also made sure that it didn't touch the table top as RF switches are sensitive to electrostatic discharge and commonly this disrupts the output. The output of the RF switch was seen at the RF1 port, in the oscilloscope. See Fig 1 for RF switch setup. 

The input was provided as mentioned in the table

3. The control terminal behaves as a TTL(Transistor-Transistor logic) and can switch on or the output depending on their state. The logic of the control for RF1 is as follows:

Although there are 3 controls for this switch, we can leave control 3 unconnected. 

4. RF driver : First I checked the output of RF driver. The RF driver is given a 9.71 V tuning voltage (to make 80Mhz for the AOM) and 1 V input. The output of 80 Mhz frequency is as seen in Fig 2. (When I made the setting of oscilloscope to 50 Ohm impedance, I couldn't change the scale of channel beyond 1 V and so couldn't see the signal in the frame. When I tried the autoset setting, it made the impedance to 1MOhm. This is the only oscilloscope I had which had 50Ohm option but uses a floppy disk and so I had to use the 1MOhm impedance option)

5. RF switch : The output of RF switch is as expected, and is the same as the RF driver output. See Fig 3 for the ON state and Fig 4 for the OFF state. 

Next Step:

1. Connect the output of the switch to AOM

2. Lock the cavity and measure the ringdown appropriately. 

Items Borrowed: I took 3 cables and a 50Ohm terminator required for this setup from elec shop. I have returned the multimeter that I borrowed previously to elec shop. I apologize for any inconvenience I may have caused.

[Shalika, Marc]

Objective: Recover PLL alignment

Details:

1. We tuned the mirror alignment to increase the main laser and auxiliary laser coupling. For AUX1 and AUX2 the mirrors circled in red and blue in Fig 1 were tuned respectively. (The output of the PD was observed in the oscilloscope but it was fluctuating a lot and so the tuning was done with all lights off)

2. The coupling status is as follows 

3. We did face trouble during tuning optics for AUX1. One possibility could be that after we changed EOM in the main laser path, the new EOM introduced a change in beam shape due to its wedge shape, and now it's difficult to focus the beam into the fiber. 

4. Also, HWP was kept before mirrors (circled in red in Fig1) to investigate polarization issues. The effect of its rotation was not observed at the AUX1 output and left the output unchanged. 

Next Step:

1. Soldering the amplifiers for the DDS3 board and doing connections. 

[Marc, Takahashi-san]

There is a clean booth close to BS chamber hosting a small optical table and microscope.

As it is too close to the LC optical table, we moved it on the other side of the vacuum tube.

The several TAMA optics (BS, TM) were placed on the bottom left drawer of the large dessicator. Note that one TM has several chips..

There were also 2 boxes that are now in the west arm.

[Marc, Nishino, Shalika]

Today's goal was initially to recover the 2 PLLs alignment.

However, we found out that one of the newly installed steering mirror to recover the beam tilt because of the wedge of the new EOM was quite miscentered.

We realigned this mirror but then also had to realign several optics afterwards..

This meant that both PLL and IRMC alignment got destroyed.

For the PLL part, we tried to recover the coupling of Main Laser to the fiber as in elog 1200.

Luckily, we still had transmitted beam after the single mode fiber.

After some tweaking we got:

[Shalika, Marc]

Overview: The Labview interface for the temperature controller, voltage controller, and polarimeter camera has been completed. 

1. For temperature controller: We can switch on/off, enable/disable channel, set temperature, choose the sensor type to be interfaced, and various other kinds of features can be accessed. (see Fig 1 and Pdf 1 attached for the front and back panels respectively)

2. Fopr voltage controller: we can switch on/off, enable/disable, set desired voltage, and sweep for any number of iterations desired. For the birefringence characterization of the KAGRA size sample, we might need more than 100000 iterations. and it can be done easily. (see Fig 2 and Pdf 3 attached for the front and back panels respectively)

3. For the Camera, we can see the azimuth, ellipticity, and power of the beam incident. (We would like to integrate the Poincare sphere, and visuals of the Polarization axis to make it better and the same as Thorlabs software. (see Fig 3 and Pdf 3 attached for the front and back panels respectively)

For all the 3 Interfaces, almost all the features offered by Thorlabs software have been integrated with the new labview interface. 

Next Step:

1. Integrate all the VI in one. Also, the GUI can be resized or components can be moved for feasibility during the experiment. 

2. Try to take real-time measurements.

[Marc, Shalika]

We followed the wiki procedure to relock MZ and GRMC.

First, we tuned the HPD bias to minimize the MZ reflection.

In this configuration we now have MZ incident power 200 mW, reflected power 8mW and transmitted power 180 mW (MZ offset is 4.2V).

We locked the GRMC and measurement transmitted power 105mW.

Then, we had to reduce the gain of both MZ and GRMC servo to 2 to lock the 2 cavities).

[Marc, Shalika]

We relocked SHG without issue.

We changed the HPD bias of MZ from 3.4V to 6.2V to minimize the reflected power (now looking like some 4th order mode).

We tuned a bit the steering mirror just after the MZ to maximize GRMC transmission.

We blocked  one arm of the MZ and realigned the beam into the GRMC with the previous steering mirror and one mirror of the MZ.

Then, we blocked the other arm of the MZ and realigned this path with the other MZ mirror.

We minimized the polarization peak by rotating the HWP before the MZ from 218.5 deg to 219.5 deg.

We measured the mode-matching of arm 1 and 2 to be 81.9% and 86.1%, respectively.

We modified DDS2 config so that DAC 0,1,2 provide 88.3 MHz.

We tuned the phase of DAC 2 (GRMC demod) from 125 deg to 165 deg to recover the expected error signal shape. The peak to peak value is about 114mV while it was 10 times larger in the past.

Then, we were able to lock GRMC.

However, we had some issues to lock the MZ : when we connect the servo out of MZ to the HPD of MZ we can hear some sound and clear misalignment..

Related curves will be uploaded soon

It can be seen in elog 3127 that our polarization camera can only measure polarization ellipticity or retardance between -pi/4 to pi/4.

However, our LC should be able to provide larger retardance meaning that we have to unwrap our retardance measurement.

To reproduce the LC datasheet, we can use the following procedure :

- find the max and min of retardance

- invert all values above the min

- add 90 deg to values below the max (plus a small offset maybe because our LC are uncompensated, here it is -7 deg)

- shift the value below the max compare to the max value using retardance := 2 * retardance_max - retardance

The results for the measurement with various LC temperature is in fig 1.

While the shape and the behavior (smaller retardance with larger temperature) seems correct, we only have about half of the retardance measured by Thorlabs..

[Marc, Michael, Shalika]

We installed the power meter in reflection of the IRMC.

We scanned the IRMC and found about 90% mode-matching. By just acting on the last steering mirror before the PBS, we could recover 97% mode-matching (see fig 1 and 2 for zoom on HOM).

The remaining HOM might be due to the beam size change induced by the new EOM + longer optical path to recover the tilt induced by the EOM wedge.

Then, we tried to lock the IRMC but most of the time we locked on this HOM.

We had a look at the error signal and it was not great (fig 3) so we tuned the phase of DDS1 DAC0 ie the phase of EOM for SHG/IRMC to 120 deg.

By reducing by 90deg we got a reasonable error signal (fig 4).

On the other hand, this gave a bad error signal for SHG (fig 5) which is recovered with phase of 120 deg (fig 6)

The reason could be that all cables from DDS were disconnected so we might have reconnected cables with incorrect length.

To compensate for that we need to add a cable with length : c/fEOM*90/360 ie about 85 cm.

[Marc, Michael, Shalika, Yuhang]

First we changed the DDS1 DAC0 and 1 frequency from 88 to 88.3 MHz (resonant freq of the EOM).

Then, we installed a powermeter in transmission of the SHG (as in elog 3091).

We scanned the SHG and tuned its alignment. In the end, we achieved about 91% mode-matching.

Then, we locked the SHG on TEM00 and monitored the reflected green power after the FI as a function of SHG temperature (see fig2).

The optimal temperature is now 3.16 kOhm which gives us about 280mW of green power.

We did same power budget as in elog 3091 :

Incident IR (measured before dichroic): 685 mW
Reflected green (measured before FI): 288 mW
Reflected IR (measured before ND2 on photodetector): 144 mW.

This gives us a slightly better SHG efficiency at around 40.2%.

It seems that we recovered the usual situation.

Last step for SHG is to install a PD in transmission to have the automatic lock of the servo working.