Marc, Yuhang

Today we went to check the GPS receiver.

First we went to the roof of the central building.

We have 4 cables going there : 1 for several environmental sigals (wind, rain, pressure, ...), 2 for radio signals and 1 for the GPS (see fig 1).

We powered on the GPS receiver but only the power led turned on.

From the user guide we also expect the led ''GPS lock' to turn on if the receiver reference clock is phase-locked to the gps one.

We monitor few output channels with an oscilloscope and results are shown in fig 2 to 4.

1 pps output delivers a continuous signal with offset of 144 mV (fig 2)

10 MHz output delivers a 10 MHz signal (fig 3)

Irig-B output delivers a modulated signal with especially the signal at 1 KHz together with time-varying sidebands. (fig 4)

We need to  check if the 1 pps signal should be generated by the GPS receiver or by the Irig-b card.

We let the GPS turn on to check if it requires time to phase-lock with GPS satellites.

Aritomi, Marc

Today, after the last absorption measurement of the annealed AZTEC sample (entry to follow), we prepared the switch to birefringence measurement.

We decreased the pump laser power to about 0.1 W and installed the 2 blades on the translation stage.

We measured the distance between the blade and the last steering mirror as 17.3 cm and 18.1 cm for the horizontal and vertical one respectively.

We reinstalled the steering mirror just after the last lens on the pump beam.

To reach normal incidence of the pump beam our plan is to do some vertical/horizontal scans at various z positions (eg 20 60 and 120 mm) to check the position of the beam on the last steering mirror.

Then we place the razor blade at the furthest position (120 mm) at the vertical/horizontal position at the level of the last steering mirror.

After some back and forth we reached 0.024 deg and 0.003 deg incidence angle for horizontal and vertical.

We'll start the birefringence measurements tomorrow.

Michael and Yuhang

We performed a ringdown measurement of OPO after replacing the OPO transmission PDA36A2 detector with PDA05CF2. The PDA05CF2 detector is made from InGaAs and has a bandwidth of 150MHz. (rise time is 0.35/150MHz = 2.3ns)

After this replacement, we locked OPO and use RF switch to turn on-off AOM to have OPO on-off resonance. We got a "ring down" of OPO transmission as the attached figure.

We can see that this is not an exponential decay. In addition, the observed decay time is much smaller than what is calculated in elog2784 of around 4us. So I double checked my calculation, and I found I made a mistake in the past about decay time. The ring down time of OPO should be 7.7ns, which is a number we are not able to measure. The limitation of this measurement comes from RF switch and AOM. The RF switch (M3SWA-2-50DRB+) has fall time of 4.6ns. The AOM rise time is 0.66*(beam diameter)/(acoustic velocity of TeO2 4200m/s), to achieve AOM rise time of 0.77ns, the beam diameter needs to be 4.9 um. Therefore, we are currently limited by the switch off time of AOM.

The fall time in the attached figure is 66ns, which is the time signal goes from 90% to 10%. Considering the current beam diameter inside AOM, the rise time of AOM is 0.66*550e-6/4200 = 86ns. There is a discrepancy of 20ns.

Note: the AOM fall time and rise time are limited by the transit time of the acoustic wave propagation across the optical beam. Thus fall time should be the same with rise time.

Michael and Yuhang

When aligning OPO, we found that the TEM00 mode position has quite obvious drift on the oscilloscope time axis. Therefore, we monitored this effect for about 10min for the old OPO in TAMA. This starting and end positions of this drift are marked by blue curcouses 'a' and 'b', as shown in the attached figure. According to this drift, we estimate the OPO optical length drift assuming the OPO optical length drift is the only cause. Also the estimation here is just a rough estimation.

1. From the attached figure, we can see that the drift of TEM00 corresponds to a time of 2.686ms.

2. The scan speed of triangular ramp signal is 0.088 V/ms. Therefore, the TEM00 movement corresponds to a PZT driving voltage change of 0.236 V.

3. We have a calibration of PZT driving coefficient of around 1.2GHz/V. So the TEM00 movement corresponds to 284 MHz. This corresponds to a length of lambda/2 * (284/4000) = 38 nm. Here lambda is 1064nm, 4000 is FSR of OPO (4GHz).

4. If we assume the cavity length drift has the same speed, we have a cavity length change of 2.3 um in 10hours. 

This number seems a bit large for me since the PZT seems to be able to control only 3um according to a datasheet in our wiki.

Marc and Yuhang

An earthquake happened around 23:48 16/03/2022 JST with earthquake center located at Miyagi/Fukushima Ken (M7.3). The earthquake in Tokyo was observed to be M3.

A check of suspended mirror coil-magnet actuators is necessary. The check relies on the use of coil-magnet actuators and oplev. Since earthquake caused mirror motion change, we need to check firstly if the suspension is still around the nominal positon relative to oplev. If not, we need to bring mirror back to the nominal positon relative to oplev so that we can use oplev to do the further check of coil-magnet actuator.

The PR oplev was found to be still good for oplev. However, Input oplev's laser beam is completely out of PSD as shown in the first attached photo. End oplev's laser beam is in PSD, but it was too far from PSD center. BS oplev doesn't work due to the broken PSD.

Therefore, we moved input mirror picomotor to recover its position for oplev. We centered end mirror oplev's laser beam as well. During the walk to end station, we check the vacuum level measured at 4 positions from TAMA center, South arm beginning, South arm center until South arm end. All vacuum levels are consistent with nominal values and no strange sound was found for vacuum pumps.

We restarted DGS system to solve the test-time out issue, which prevents us from taking oplev noise spectrum measurement. Although we took snapshot for medm setting, we couldn't recover medm setting afterwards. We suspected that this was caused by the 'diskfull' issue. We should pay attention next time.

After all these work, we ran the automatic code as reported in elog2836. We found weird response for Input H4, End H1, and End H3 as the attached figures 2,3,4. Note that the blue and red curves are reference oplev spectrum. The other two curves (green and brown) have difference spectrum and almost no coherence. Therefore, we suspect Input_H4, End_H1, End_H3 magnets may fall down.

This morning I wanted to start a new measurement on the XZ plane.

Before any measurements we need to check the input power so it is convenient to just move the mirror outside the beam.

I made a mistake in the motion and actually could see that the beam was cut somewhere.

I started the measurement to try to see features on the mirror surface but it was not so much conclusive so I went to TAMA and found out that the bottom left of the holder got a little burnt (see picture 1).

After discussing with Matteo, I decrease the pump laser power, closed the shutter and removed the holder from the translation stage.

I inspected the surfaces of the mirror with the strong green light and it seemed to be not too dirty but to avoid any doubts on the measurements results I cleaned them with alcohool, lens cleaning tissue and ion gun.

I also cleaned the holder (see picture 2) and removed dust from the translation stage.

I added more stringent limits on Zaber that should be changed before KAGRA size sample measurements.

The mirror is now reinstalled and I'm waiting for the pump laser power to stabilize before restarting measurements.

Katsuki, Marc

We found the GPS receiver on the left of the network rack in the storage room (see attached pictures).

We tried to follow the black cable that seems to indeed go to the roof of the central building.

While LIGO recommends to have the cable between the GPS antenna and receiver as short as possible, our cable is really long so we should be able to have our GPS receiver on the rack close to the input mirror.

The GPS receiver user guide is accesible here : www.leapsecond.com/museum/hp59551a/097-59551-02-iss-1.pdf

Yuhang, Marc, Michael

We took inventory for the computer equipment located at TAMA that is required for the DGS upgrade. Specifically, we would like to increase the number of ADC/DAC channels to accommodate new QPDs that should be installed. However, our current computer system is full, and we would like to install new frontend PCs to operate the additional channels. By adding a new PC(s), we have to go from standalone to network configuration. In a network configuration, synchronised data acquisition is achieved using the Dolphin network equipment. A summary of the principles and components of the ALIGO timing system is written by Yoichi Aso et al. "Advanced LIGO Timing System: Final Design" LIGO-T070173-00-D

1000BASE-LX/LH SFP 1310nm 10km Transceiver Module: 2 boxes of optical fiber transceivers (USB-stick sized optical to electronic transceiver)

24-Port Gigabit PoE+ Managed Switch with 2 1Gb Combo and 2 10Gb SFP Uplinks, 440W: 1 unit. Large black chassis. Model number S3410-24TS-P.

24-Port Gigabit L2+ Stackable Managed Switch with 4 10Gb SFP+: 3 units. Large grey chassis. Model number S3910-24TS.

48-port Gigabit PoE+ Managed Switch with 4 SFP+, 400W: 2 units. Large grey/black chassis. Model number S3400-48T4SP. 

Thick 5m optical fiber cables: 3 units. SA-444L-5M-S2.

Digital to analogue converter: 2 units of PCI express cards. PCIe-16AO16-16-F0-DF.

Analogue to digital converter: 2 units of PCI express cards. PCIe-16AI64SSC-64-50M.

Dolphin PCIe Gen 3 PXH830 non-transparent bridge adapter card: 3 units (in nondescript brown boxes)

Adnaco PCIe epxansion host card and chassis: One unit. ADNACO-S2B-01-000 card and Adnaco-C2B chassis. 

Myricom card: One unit. Used for ultrafast communication of frontend PC to data concentrator.

Senetem CAT.7 flat cable, 10 Gb LAN: 6 units 2m, 6 units 5m.

10m thin optical cable: Unlabelled. Blue colour.

IRIG cables: One box. Used for communication from GPS receiver to timing system to retreive the absolute GPS time.

DGS Timing Card: One unit. Spectracom, located in bag "DGS Timing Card". I'm guessing it coordinates GPS absolute time with the 1PPS GPS synchronisation signal.

Dolphin MXS824 24 port PCIe Gen 3 switch: One unit. Large black chassis.

-- 

Some already installed computer units:

1. Model # CSP-30EGSR4, SN CS6279 (next to PR tank)
2. Model # CSP-32XES, SN CS6257 (rack next to south arm, upper left)
3. Model # CSP-30EGSR4, SN CS6285 (rack next to south arm, upper right)
4. Model # CSP-38XQDR4, SN CS6278 (rack next to south arm, middle)
5. Model # 813M-3, SN BT0-2947105-001 (rack next to south arm, bottom)

Katsuki, Marc

This morning we checked again the pump beam profile to make sure the earthquake did not affect its properties.

It was fine so we checked the surface calibration (R = 16.64 at z = 35 mm and z_IU = 68mm) and bulk calibration (R_bulk = 0.6909 cm/W).

We inspected the sample with the strong green light and cleaned it with the ion gun.

We checked the centering and got X_center = 326.99mm and Y_center = 122.709 mm.

We increased the laser power to about 7.5 W and did a long z scan from which we got the 2 surfaces at z = 41.34 mm and 76.02mm making Z_center = 58.68 mm.

We could also estimate the absorption to be about 70 ppm/cm (roughly same as previous measurement).

We started a XY absorption measurement at Z_center with 0.25 mm step size, 0.5s waiting time and 10 order average/median filters that will last for about 16 h 30.

Today we removed the SHINKOSHA 7 with Yuhang.

Actually Yuhang pointed out that our technique to remove this heavy sample generates quite strong impact on the injection breadboard that could be one of the reason for the pump beam shape change.

I checked the surface reference sample and got R = 16.50 /W instead of the previous R = 16.91 /W.

I suspected that the difference was again arising because of a change in the pump beam size.

I installed the razor blade cutting the beam vertically and got a pump beam waist of 35 um at z = 58.16 mm (instead of the previous 35.4 um at 57.6 mm) as reported in figure 1.

The plan is to install the surface reference sample 0.56 mm further away (ie at z = 35.56 mm).

Thank you!

I also finished the last 2 measurements in between the 3 previous measurements that are attached to this entry.

Great achievement, Thank you!

Once the new data are ready, we can finish up the paper

Yuhang and Michael

We attempted to measure the OPO ringdown using the new RF switch described in 2865, 2866. 

Figure 1 shows the result of the ringdown for the transmission (red) and reflection (blue). The behaviour is a bit strange, so we haven't analysed in detail yet. The reflection ringdown is too fast, while the transmission ringdown starts off too fast and then becomes too slow. As a reminder, we are expecting about 3-4 µs ringdown from calculation. The RF switch is controlled by a 5V 1Hz square wave (i.e. digital logic on/off). Figure 2 shows the result of switching off the switch control manually. The electrical contact bounces and causes the switch to rapidly turn on and off. Normally this result doesn't seem very useful, but even here we can see the reflection ringdown being too fast and the transmission ring down being too slow.

Current mode matching status is:
34 mV TEM00
8.6 mV HOM
5.1 mV HOM
3.9 mV noise floor

We also attempted to measure the optomechanical transfer function of the cavity. We used Mokulab's Frequency Response Analyzer, taking Mokulab IN1/IN2, with:
Mokulab IN1: Taken from T connected to Source A of SR560. This is the input of the servo before noise injection.
Noise injection: Injected at Source B of SR560, which applies a low pass filter to A - B.
Mokulab IN2: Taken from T connected to 50 Ohm out of SR560 (before PZT high voltage driver). This is the output of servo going to the plant. Thus, IN1/IN2 should be the TF of the plant.
The result is shown in Figure 3. However, it is not very meaningful. The error signal is probably too high versus the noise injection for frequency response analysis.

Yuhang and Michael

We came back to the OPO setup and found the beam was quite misaligned. We could not really reduce the presence of a certain higher order mode (15 mV vs 65 mV total mode power). The PZT also seemed to be drifting a lot even though the temperature was controlled.

We finished the first 3 measurements taken at the same positions as in Caltech ie at the mirror center, 10 mm after the first surface and 10 mm before the second one.

The results are attached to this entry and compatible with their measurements (and therefore also with Manuel's ones).

So we started absorption measurements in between these positions to get more data for the integrated map along z.

Today I reinstalled the reference samples and checked the proper positions.

With the surface sample I got the crossing point at z = 35 mm (maximizes AC) and z_IU = 68 mm (maximizes AC/DC).

I measured R_surface = 16.91/W and R_bulk = 0.6212 cm/W.

I installed back SHINKOSHA 7 and did a long z scan.

In the attached figure you can see the comparison between various signals for the previous long z scan (red) and current situation (black) taken at the mirror X and Y centers.

Note that the z axis has been shifted for the red using the surfaces signal and then interpolated to the new measurement (step size of 0.05 mm instead of 0.1 mm).

As expected, the absorption is larger now and it seems that we have something like at least a factor 1.4 increase.

However, I was expecting to see a somehow constant increase but this is not the case..

I started a XY absorption measurement at the mirror center (ie X = 399.08mm, Y = 122.175 mm and Z = 71.625 mm)

Katsuki, Marc

This is a summary of these past days activities.

As reported in entry 2863 we found out that the pump beam was larger than expected (48.5 um instead of 36 um).

In summary we had to act on the two lenses on the pump beam path to recover the good beam size and position following Jammt simulations.

These 2 lenses are now about 1 cm closer to the laser source.

During this realignment we also checked the the probe beam size to have a reference waist position.

We found out that it is not feasible to use the absorption DC photodiode together with the razor blade because there is scattering when we start to cut the beam that creates a spikes in the data and prevent a good fit of the data.

Furthermore, the probe beam is really large on this photodiode and is really astigmatic when setting up the imaging unit translation stage at z_IU = 0 mm (ie farthest from the translation stage).

In the end, we installed a power-meter in between the imaging unit lens and sphere and could get good data.

We also found out that a good step size for the translation stage is 20 um as it allows to get good enough resolution while not taking too long.

The attached figure reports the beam profiles of probe beam in vertical and pump beam in both vertical and horizontal directions.

The z axis is the same as Manuel's measurement (see elog 1089) ie the 0 mm is at 75mm from the breadboard.

We had to tweak a bit the tilt of the lenses to minimize the pump beam astigmatism.

We recovered the expected pump beam waist size and position so we will now switch to absorption measurement.

Note that probe beam power is 2.5 mW and pump beam power was about 150 mW for this measurement.

Nishino

I measured the beam profile between L2 and L3. I got better results than the previous one (see 2856).

*starting position is 25 mm away from L2.

Taking the first 30ns of measurement data, I did a FFT analysis of the data and got a power spectrum density (PSD). Then the time span is shifted by 1ns several times to get the PSD evolution. In total, the 200ns data is shifted by 170 times to get the signal PSD change as a function of time. This is shown in attached figure one.

The FFT has a bandwidth of 33MHz (since I used 30ns to make a FFT). Because the RF signal has a frequency of 110MHz, I took the frequency span of 99-132MHz to check the amplitude of RF switch output.

From this analysis, the fall time, which is the time that signal drops from 90% to 10%, is 10.6ns.

In addition, I also put a time-frequency-amplitude plot of this signal.

Pierre Prat (remote), Yuhang and Michael

We received the Minicircuits M3SWA-2-50DRB+ absorptive RF switch evaluation board. Nominally, it has a fall time of 4.6 ns, well within bounds of what we want (400 ns). In this case, the rise and fall times have been specified by the manufacturer as the time it takes to go from 10% to 90% of the peak voltage and vice versa. The circuit can accept high input power > 24 dBm at 100 MHz. It is powered by -5/+5 V supply. The switch is activated/deactiveated by a TTL (transistor-transistor logic) control signal. In short, voltages in a certain low threshold (0-0.8 V) are considered "OFF" and in a certain high threshold (2.1-5 V) are considered "ON". In this case, we can just use a square wave oscillating between 0 and 5 V, and then trigger the oscilloscope to follow the rise/fall of the RF signal.

Chip manual: https://www.minicircuits.com/pdfs/M3SWA-2-50DRB+.pdf
Evaluation board diagram: https://www.minicircuits.com/pcb/WTB-M3SWA250DRB+_P02.pdf
 

-- Test --

The RF switch was tested in the filter cavity clean room using the already present oscilloscope, function generator and RF amplifier(s). We brought a DC power supply to send -5V/+5V to power the RF switch, as well as a Tektronix AFG320 function generator to provide the control signal to the RF switch (0 to 5V square wave, checked at 1 Hz and 12 kHz). Both of these were tested first to make sure they give the required voltage and square wave signal.

A 500 MHz RF signal was sent from the filter cavity function generator to the switch -> RF amplifier -> oscilloscope. A 20 dB attenuator with 50 Ohm impedance was connected to the oscilloscope to prevent back reflection. Unfortunately, the first RF amplifier (Minicircuits ZHL2) we were using stopped outputting. We did take care to say the order in which you should make connections with the RF amplifier. I hope it is not permanently broken... 

-- Data --

The figure shows the fall time when a 1 Hz square wave is sent to the TTL port of the switch (rise time figure pending). The data is a bit low resolution. The lower half of the figure shows a zoom in of the timescale and indicates 10% of Vpk. This measurement doesn't seem very accurate, but regardless, the fall time is well below the target of 400 ns.

With these results, we moved the RF switch and the necessary electronics to the ATC cleanroom.