Stable lock of the X End green laser was recovered.

- The biggest issue was that the laser PZT input had been terminated with a 50ohm at the laser head. (See Attachment 1: The terminator has already been removed in the photo.) Since the PZT output of the servo box (output impedance 10Ohm) goes through 680Ohm at the summing node for the modulation, the PZT output was attenuated by a factor of 15. This made the required servo gain for locking more than the box could deliver. More importantly, the PZT range (in terms of the laser frequency) was also limited. Momentary locks were still possible with the reduced range and gain. However, the actuation signal hit the rail within a few seconds because of the pendulum motion.

Once the terminator was removed from the head, the Xarm was locked with the green laser like a charm.

- On the way to the resolution, I had to go through the full scrutinization of the loop components one by one. Here is the record of the findings:

• Inspected the green Refl PD (Thorlabs PFA36A). The gain setting of the PD was 40 dB, and the unlocked output voltage was 10.8 V. This is not only very close to saturation, but also the bandwidth drops below the modulation frequency (150 kHz according to Thorlabs' manual). The gain was changed to 20dB. This made the unlocked PD output to be 1.08V and the BW was expected to be 1MHz.
   
• Checked the LO setting. The box has a label saying "LO 7dBm". The function generator setting of "0.66 Vrms" resulted in 7.0dBm at the mixer LO input. So this number is used. Exactly the same amount goes to the PZT summing node.
   
• Checked the mod freq. The PDH error signal amplitude was maximized at 278.5kHz (mixer output observed with 50Ohm: 46.0mV), however, the signal looked distorted from the text-book shape of the PDF error. This means that the demod phase was not optimized.
  The mod freq of 287.5kHz made the PDH error signal look better while the response was weaker (mixer out: 31.2mV). It turned out that the cavity locking didn't like these mod freq between 280kHz~290kHz. The momentary lock stretches showed a lot of quasi-sinusoidal fluctuation ~600Hz in the error and transmission signals. Instead, the modulation of 210.5kHz was used. This made the error signal during lock stretches clean and tight. 
   
• Box inspection: Checked the signal ratio between the error in and the error mon. The monitor gain seemed x20~x21. The PZT output and the PZT mon had identical gains. The transfer function of the box was measured with the gain knob changed from 0.00 to 7.00 where the transfer function started to get distorted with the given input. The gain was increased by 5dB/turn (i.e., 1 turn increases the gain by 5dB). ? It does not match with the info on the schematic and the datasheet? Anyways, the gain knob is working fine.
   
• To resurrect the SLOW THERMAL servo, the monitor channels were connected to the DAQ interface. The existing slow channel servo/setting worked fine, wh
   
• Usual caution: a slight touch to the satellite amp caused the UR OSEM PD completely black out. It means that just your presence at the X end can make some changes to the suspension.
   

After the XAUX - XARM lock was recovered the C1:ALS-TRX_GAIN was set from 0.002 to 0.0006 to normalize the green transmission to 1 when the cavity is aligned. This situation was verified with YAUX as well. The green transmissions are now normalized to 1 when both arm cavities are aligned.

After this I took a reference ALS noise spectra (Attachment #1). The XALS rms noise is ~ 100 Hz (which is great compared to previous reference of > 250 Hz), while the YALS is slightly worse at high frequency but the rms is comparable to previous references (~ 250 Hz). This is somewhat encouraging for our future PRFPMI lock acquistions.

I measured the OLTF of the XAUX-PDH loop [Attachment 1] now that the green laser is stably locking. I injected an excitation (100mVpp) at the error point of the loop using a Moku:Go. The excitation was summed with the PDH error signal (alpha) using an SR560, and the summed signal (beta) was sent to the PDH servo. (The Moku excitation was buffered with another SR560.) The transfer function beta/alpha was measured on the Moku. 

The loop has a UGF of 26.3 kHz, and a phase margin of ~25º (using 1/1-OLG convention).

Next steps:

- Replace PDH servo demod + controller with Moku:Go lock-in amplifier (ensure loop shape is maintained)

- Deploy digital filters to further increase loop bandwidth/phase margin

I calibrated the moku phasemeter setup for reading beatnote fluctuations today. The calibration is referred to the DFD output (not including the phase tracker) channels by using the measurement made by gautam in 40m/14981.

Measurement setup

• Set marconi to 40 MHz carrier, FM1 sine deviation of amplitude 2000 kHz (we expect maximum beatnote fluctuation of ~1.8 kHz for 50 pm length modulation in the arm length) at 145 Hz.
• The output of amrconi is splitted, one half going to DFD for BEATY, one half going to Moku Phasemeter input 1.
• Moku phase meter is set with following settings:
  • Input1:
    • Frequency : Auto
    • Bandwidth: 1 MHz
    • Coupling: AC
    • Impedance: 50 Ohms
    • Range: 400 mVpp
  • Output1:
    • Signal Freq offset
    • Scaling: 1 mV/Hz
    • Invert: Off
    • Offset: 0
    • Range: 10 Vpp
• Measurement taken from 1365442502 to 1365444003

Analysis

• Read channels C1:ALS-BEATY_FINE_I_OUT C1:ALS-BEATY_FINE_Q_OUT C1:ALS-BEATY_MOKU_PHASE_IN1 for 500s.
• Used np.arctan2(Q, I) to read DFD phase output. Multiplied it by 1e6/70.973 to convert it into Hz using DFd calibration by Gautam in 40m/14981. This measurement brings in 340 ppm of uncertainty in the measurement.
• Demodulated the phase at 145 Hz to get the signal sent by Marconi. Blue trace in the attachment is this signal.
• Demodulated Moku phase output at 145 Hz and calculated the calibration constant required to match the ampltiude with 400second averaged DFD output.
  Calibration constant came out to be: 0.2953 +/- 0.0001
• Multiplied the calibration constant to moku phase output. This is the orange curve in the attachment.
• With this method, we get 0.035% uncertainty on phase calibration from Moku.
• We can now use moku phasemeter for calibration measurements as the pahse tracker gain is not high enough for calibration lines above 200 Hz.

Ditto of 40m/17543 but for XBEAT >> Calibrated to 0.3061 +/- 0.0001

An interesting thing to look at is the ALS out of loop spectra using our Moku DPLL. Attachment #1 shows the calibrated noise spectra in Hz/rtHz of both XBEAT and YBEAT as taken by the Moku phasemeter when the ARM cavities are locked to PSL using POX/POY. A great improvement is noted at lower frequencies (almost an order of magnitude for Y, over an order of magnitude for X) and some residual seismic noise (between ARMs and IMC) is noticeable! At higher frequencies, the suppressed laser frequency noises are close to their former references.

However this is only great news for our ALS calibration scheme, as the DPLL range is limited and may not be useful for the usual CARM offset reduction using ALS.

The rms fluctutations for the ALS beatnotes using the Moku Phasemeter have dropped below the 100 Hz floor! We now have 50-55 Hz (before we had 200 to 300 Hz)

Took some ITM/ETM single arm calibration data using Moku Phasemeter ALS>

• ITM gpstimes = [1365471104 to 1365471582]
• ETM gpstimes = [1365472286 to 1365473448]

I then took some ITM actuation calibration data using the Michelson fringe. For this I lock MICH and turn on all five lines using AS55_Q >

• ITMX gpstimes = [1365474355 to 1365474788]
• ITMY gpstimes = [1365474816 to 1365475268]
• Free swing = [1365475309 to 1365475535] (to get the fringe amplitude)

The ETM actuation calibration at these frequencies can be transferred using the POX/POY error signals and the ITM calibration from the gpstimes above. This should allow us a back to back cross-calibration comparison for arm cavities. Full analysis to follow this entry.

Finally, please take note of the area around the LSC rack! The temporary Moku phase meter calibration and setup referenced above are still a bit in the way. See Attachment #2

I was able to get this accelerometer going for the next Lab tours. I want to get this guy up on a big screen to give people a nice "wow". I found this accelerometer on the Y end cabinet and there is 1 more available if anyone needs it at 40m. It is a Brüel & Kjær 8318. It contains a PZT so there is no need to input a signal. The accelerometer seemed to only put out roughly 2 mV max, so i had to amplify with an SR560 to get a good looking signal. 

RXA: link to Manual

[Anchal, Yuta]

We have repeated LO phase noise measurement done in elog 40m/17511.
Method we took was the same, but this time, we used (1+G)*[C1:HPC-LO_PHASE_IN1]/[optical gain] to estimate the free-running noise, instead of using [C1:HPC-LO_PHASE_OUT] multiplied by LO1 actuator gain.
We confirmed that both method agrees down to ~ 10 Hz (at lower frequencies, OLTF measurement is not robust; we used interpolated measured OLTF (Attachment #1) for compensation).
Below is the summary of optical gains etc measured today.
Filter gains were adjusted to have UGF of 50 Hz for all.

LO_PHASE lock in ITMX single bounce
        Demod phase  Optical gain     filter gain
BH55_Q  -102.7 deg    6.9e9 counts/m  -0.34
BH44_Q  -5.7 deg     1.3e10 counts/m  -0.17

LO_PHASE lock in MICH
        Demod phase  Optical gain     filter gain
BH55_Q  -72.6 deg    8.7e8 counts/m   -4.4
BH44_Q  -27.6 deg    8.8e8 counts/m   -2.2

LO_PHASE lock in FPMI
        Demod phase  Optical gain     filter gain
BH55_Q  24.2 deg     3.7e9 counts/m   -0.67
BH44_Q  2 deg        5.3e8 counts/m   -4.4  (An order of magnitude smaller than elog 40m/17511)

The values are consistent with elog 40m/17511, except for BH44 in FPMI.
It took sometime to robustly rock LO_PHASE with BH44_Q in FPMI today.
After some alignment, offset tuning and demod phase tuning, it finally worked.
Demod phase of BH44 was tuned to have more DC signal when LO_PHASE was locked with BH55_Q, considering that BH55 and BH44 are orthogonal.
It actually created BH44_I having more amplitude (some noise?) than BH44_Q, but BH44_Q was more coherent to LO_PHASE fringe in BH55_Q.
It might be related to why we are not dark noise limited for BH44_Q, while BH55_Q is dark noise limited in FPMI, and why we cannot lock FPMI BHD with BH44.

Attached are the Rayleigh spectrograms of the error/control signal channels associated with the NN nonlinear control of IMC (pitch). The 4-hour data stretch starts at 3:45pm PDT on 4/18. The spectrograms were generated with (stride=5, fftlength=2, overlap=1). PNG images are attached for reference; the generated pdf files were too large to include here or send over email.

The Rayleigh statistic measures nongaussianity of the data.

First, simple stuff. We estimate the noise budget with total input and output noises. Later, we will break it down (ADC, DAC, whitening, dewhitening noises etc.):

We take the dark noise of AS55, REFL11 and make sure that the whitening and "unwhitening" software filters are on (attachment 1)

To convert cts to Watts we use the values from previous MICH noise budgeting for AS55:

PD_responsivity = 1e3*0.8 #V/W
ADC_TF = 3276.8 #cts/V
demod_gain = 2 #6.77 #According to https://wiki-40m.ligo.caltech.edu/Electronics/LSC_demoddulators
whitening_gain = 10**(24/20) #24 dB

We are not sure why the demod gain was chosen to be 2 and not 6.77 as in the Wiki, maybe it was chosen to match the measurements back then. The demod gain for AS55 was actually measured to be 2.4 in elog 16961.

For now, for lack of time, we use the PD responsivity and demod gain of REFL11 from the wiki:

PD_responsivity = 4.08e3*0.8 #V/W
ADC_TF = 3276.8 #cts/V
demod_gain = 4.74 #According to https://wiki-40m.ligo.caltech.edu/Electronics/LSC_demoddulators
whitening_gain = 10**(18/20) #18 dB

Using the Finesse model for PRMI (should push to git) we calculate the sensing matrix (attachment 3). We turn off the HOMs as it gives us strange results for now.

We take the output noise that was measured at the output of the BS coil driver measured in elog 16960.

Attachment 2 shows the estimated PRMI noise budget. Notice that the dark noise contribution is an order of magnitude better than MICH (elog 16984) due to PRG.

Today, we ran dither lines on the MC1,2,3 mirrors in YAW from 136598007 to 1365981967 and similarly on PIT from 1365982917 to 1365984618.

The following frequencies and amplitudes were recorded for each dither line:

The urad conversions used to calculate theta DC and AC can be found at 17481

The dither lines were then demodulated in python and the steps shown in 17516 were followed to calculate the beam offset that each dither line represented in pitch and yaw. 

The following results were found:

Attatched bellow is the power spectrums for both yaw and pitch.

[Anchal, Paco]

We updated the LSC model to use the amplitude as a normalization (analogous to what happens in OpLevs). For reference Attachment #1 shows the previous model detail, and Attachment #2 shows the updated one. We then built, restarted and ran the model to realize the phase tracker gain can now be set once and for all assuming we still have a simple integrator and 2 kHz of phase tracker bandwidth. Doing this results in the ALS residual noise shown in Attachment #3. Compared against the reference spectra, the improvement is modest but not as great as what the moku had.

We ran ITMY actuation calibration using this infrastructure; to do this we lock arm cavities to PSL, lock AUX lasers to arm cavities, turn on our five lines and read back the demodulated signals from the beatnote as it goes through DFD + phase tracker. The results are summarized in Attachment #4. This time we correctly accounted for all known sources of statistical and systematic uncertainties (including a recently  measurement of the AUX loop gain),

how about the other idea of downloading the I & Q channels and doing the analysis offline? I'm curious if its better or worse. How could the Moku possibly be better?

Another idea is to use the frequency divider and then directly digitize. I believe someone tried that a few years ago, but not sure how good it was.

I did offline analysis with the available data. We were only saving signals at 2048 Hz rate, so analysis can not be done on 1.4 kHz line. See attached plot for the difference in the two analysis.

We are aiming to prepare a realtime system deployable calibration method, that's why we were using phase tracker. Note that the calibration results with phase tracker have been compensated for any lack of gain due to phase tracke limited bandwidth, open loop gain of aux loop or remaining suppression from YARM loop despite the notches.

About the moku, we think that something is wrong in connection of moku output to ADC. We see the same cal line heights in the moku app in ipad but after going through ADC, we see about 10 times less line heights and 10 time sles noise floor too. But when we stick a marconi split between DFD and moku, we see the same results, so we are not sure what is wrong with it but it is not trustworthy. Maybe the order of magnitude noise reduciton is because of this factor of 10 that happens when it reads beatnote. To be solved in future, we will carry on with DFD for now.

Last night I took ITMY calibration data using MICH with AS55_Q. Adding that to the same plot. The error bars are probably underestimate with the MICH calibration method due to systematics not taken into account. For this measurement, MICH was locked with low UGF of 20 Hz to avoid all lines in MICH loop. Notches at the line frequencies were also put in. MICH OLTF was measured and any possible suppression of lines has been compensated for (very small). Note that error bars are present for DFD method too, but they are too small in this scale.

MICH calibration did not independently verify the higher actuation strength found by DFD methods at higher frequencies. For an ideal pendulum, the calibration constants should ahve been freqeuncy independent. It does see higher calibration constant values at 500 Hz and 1.4 kHz lines, but with a lot of noise. See attachment 2 for the calibration in real time, but this plot is bit messy. For the three lower frequency lines, DFD+Phase tracker and DFD with offline analysis match in their estimates , there is a significant mismatch at 500 kHz line and we do not have data for doing this for 1.4 kHz line.

This morning I found a helpful post in the git.ligo pages that referred to the issue that was stopping the summary pages from being submitted to the codor. In the last 3 weeks the accounting_tag ligo.dev.o3 was retired, and thus the accounting tag "ligo.dev.o3.detchar.daily.summary" which is found in both /condor/gw_daily_summary.sub and /condor/gw_daily_summary_rerun.sub was preventing the summary pages from bieng submitted.

The post referred to a running list in the nodus that points to all active accounting tags: etc/condor/accounting/valid_tags this showed that "ligo.dev.o4.detchar.daily.summary" and "ligo.dev.o5.detchar.daily.summary" were currently active tags. I then tested the files using this tag (o4) with success in starting the pages up again. I have now committed the change to the git. 

The prior 3 days of summary pages were also regenerated: 

relocate to /bin

run ./gw_daily_summary_40m --day YYYYMMDD

once finished run ./pushnodus YYYYMMDD

      (must do independently for each day)

Mayank and I worked on finalizing the plots for the beam offset from the dithering test done in 17552. Plotted in attachment 1 are the beamspot demodulated signals from MC_F_DQ which are averaged over 1 second each (blue) for YAW and PIT in MC1,2,3. The yellow line over each plot shows the 3 Hz lowpassed signal of the beamspot movement.

Additionally, we have seen no direct correlation to the WFS1 or 2 sensors due to the MC movements. This may be because the WFS display a complete signal that includes all changes in the cavity length due to the shaking of the mirrors. Thus, the signal (shown in red) of the WFS sensors will show a combined average of movement from all 3 dither lines.

It has happened multiple times today that IMC has tripped on its own. Yehonathan and I have had to come back to manually lock IMC multiple times.

Wed Apr 26 10:24:07 2023 [EDIT]

[Paco] I aligned the MC by hand, let it run locked for 30 minutes without angular controls, and then switched on the WFS loops yesterday at ~ 6 PM. IMC has been locked ever since.

I modified the analysis to correct for any affects due to Anti-Aliasing or Anti-Imaging filters, and I also found a insignificant error on how I was undoing the suppression due to MICH loop in the MICH data. I also propagated the calibration in MICH method better. Attached are the updated results. The upward swing is still present.

Also, last night, Koji and I looked into any frequency dependent deviation in sensing arm length between POY11 and BEATY_PHASE (using DFD+Phase tracker) This was done by locked the YARM to the main laser and locking YAUX to the YARM, sending excitationa at C1:SUS-ETMY_POSCAL_EXC and taking transfer function between C1:LSC-YARM_IN1 and C1:ALS-BEAT_Y_FINE_PHASE_OUT. This transfer function was flat upto about 600 Hz and the deviation from there to 2000 Hz was expected based on limited bandwidth of the phase tracker. I don't have the plot to attach, someone should redo this quick measurement to save the data.
Interestingly, the same measurement when done with  C1:LSC-DARM_IN1 in FPMI configuration did not show a flat response. This is can mean that the DARM strain relationship with the beatnote frequency deviation is not a simple constant factor and/or depends on DARM or CARM OLTFs. I leave my remarks on this project here for the baton to be picked up by others in future. I unfortunately only have this much time to contribute to FPMI calibration.

{Mayank, Paco, Yehonathan}

Dewhitening noise curves were taken using SR785+SR560 for the PRMI noise budget. One representative channel was measured at each board, suspensions were tripped before work was done. The input pins to the dewhitening boards were shorted using an exposed ribbon cable.

At each board, the measurement was taken with and without dewhitening filter on. The toggling of the dewhitening filter was done by turning on and off the SimDW filters at the coil filter screen of each suspension.

Attachment 1 summarizes the results.

ITMX dewhitening noise is much higher than the rest.

ITMY measurement turned out to be bogus since we mostly measured dark noise. The reason we made the gain so low in that measurement is that it was saturating the SR560 whenever we used gain>1.

We measured the IQ demodulation board gains for REFL11 and AS55.

To do this, we replaced the PD input on the demod board with an RF signal at near the nominal frequencies of 11.066195 MHz and 55.330975 MHz using a Marconi 2024A identical to the one which sources the PM sidebands in our PSL. Even though we matched the modulation frequencies we found the two marconis were in practice offset by ~ 3 Hz. After tuning the frequency around a bit, we managed to get them to within 450 mHz.

REFL11

We started with REFL11 IQ demod board. After sourcing 11.066198 MHz into the PD input port, we took the I and Q outputs and looked at them using an osciloscope. We measured the Vpp levels on both as well as the Marconi output. The resulting levels were

• Source = 44.4 mVpp, I = 8.8 mVpp and Q = 10.8 mVpp ==> Gains are therefore 0.19 and 0.24. The amplitude gain of this board is sqrt(0.19 ** 2 + 0.24 ** 2) = 0.153. This is in stark disagreement with the wiki. Has the wiki finally failed us?

AS55

We then moved on to AS55 IQ demod board. After sourcing 55.330975 MHz* into the PD input port, we took the I and Q outputs and looked at them using an osciloscope. We measured the Vpp levels on both as well as the Marconi output. The resulting levels were

• Source = 16.8 mVpp, I = 50.8 mVpp and Q = 56.3 mVpp ==> Gains are therefore estimated to be 3.3 and 3.7. The amplitude gain of this board is sqrt(3.3**2 + 3.7**2) = 4.74. This is in slight contrast with the previously measured gain of 2.8, but we think a factor of 2 may have been misplaced in either calculation since one typically estimates AMP = 2 * sqrt(I**2 + Q**2).

* Note that in the second test, we didn't match up the frequency, which caused I and Q outputs to have significant gains (instead of just I).

Tl;dr: Tried to replace of XEND green PDH servo controller with Moku template IIR filter, designed to match PDH servo frequency response. The green laser did not catch lock with this filter.

Attachment 1 plots the measured TF of the PDH servo controller, with boost on and the gain knob set to 7.22 (the current lock configurations). It also plots an 8th order Chebyshev type II low-pass filter, with cutoff frequency and scale chosen to best match the data. (8 was the highest order filter that could be represented by 4 second-order-sections, the maximum allowed by the Moku.) I wanted to test if the XAUX PDH lock could be maintained using this filter as the controller.

The phase of the Chebyshev II filter does not seem to be a good fit to the data, but I wanted to see how far we could get using a template filter already designed for discrete time, and with a magnitude frequency response that approximates the servo. This would bypass having to perform a bilinear transform from the s-domain to the z-domain, which can raise more complications.

The PDH error signal (mixer output) was split and sent to the Moku (input 1) and to the PDH servo input. Closing the loop with the Moku filter output, the green laser was not able to catch lock. Attachment 2 shows the Moku:Go Digital Filter Box configurations, as well as the traces comparing output of the filter and the output of the PDH servo. The red trace is the output of Moku filter, and the blue trace is the output of the PDH servo (input 2) with the loop open (nothing feeding back to laser PZT). The input gain of the filter module was chosen to match the amplitudes of the two control signals. Qualitatively, the filter output contains higher frequency components and preserves the odd polarity of the PDH error signal, compared to the servo output. 

I then tried to directly fit the PDH servo TF data. I fit the (analog) poles and zeros of the TF using vectfit. In theory, using a bilinear transform can convert the analog zpk TF to digital zpk, with some frequency pre-warping required. However, vectfit did not return a "normal" transfer function, defined as having at least as many poles as zeros. This caused the bilinear transform to fail.

Next, I will need to use a different fitting package (perhaps IIRrational) to obtain a nicer TF fit, in normal form. Then I can attemp the bilinear transform, confirm it preserves the desired frequency response, and test it out with the Moku:Go.

I included the output electronic noises into the PRMI carrier noise budget (attachment 1).

The coil driver noise was calculated using the Johnson noises of the coil driver resistor:

PRM 430 ohm

BS 100 ohm

ITMX/Y 400 ohm

For the dewhitening noises I use the measurements from yesterday. As expected fro yesterday's measurements, the ITMX dewhitening noise is dominating. For the coil driver gain I use the recently measured actuation calibration (elog   17522 ) to extract it. I find that these gain values:

PRM 1.009

BS 1.333

ITMX/Y 0.24

For the DAC noise I assume 1uV/sqrtHz and use the simDW filters from the coil outputs MEDM screens as the DW filters TFs.

Next:

1. Break down input noises.

2. Measure how much light is reaching REFL11 to correct the sensing matrix and get the right shot noise.

[Paco, Koji, Mayank]

We replaced ITMX Oplev HeNe laser after last one died.

• We noticed the ITMX HeNe laser died, so we couldn't have Oplev control. With Koji's help, we located a laser tube which had no labels and a sticker up front so we assumed this is new. We removed the faulty tube from the ITMX table and tested it using the display control unit (labeled 1). The laser head was 1103, and weirdly it turned on. We also located some control units in the Oplev cabinet which had been labeled "working with Display" from 2022.
• After testing a few combinations we realized CARE NEEDS TO BE TAKEN TO NOTICE THE INTERLOCK SWITCH BEFORE TESTING ANY PAIR> it turned out our old controller was ok, but the interlock switch had fallen. We also noted the two spare units labeled "working with Display" were indeed operating correctly they just needed their interlock switches. Surprisingly the "new" laser head we planned to use as a spare is DEAD.
• After assuming the old head was working and its controller just needed the interlock, we reinstalled it on the table. Unfortunately, after realigning the OPLEV beam we saw the SUM drop from 1500 counts (usually 15000) to 0. So the laser head was indeed faulty! + WE HAVE NO MORE 1103 SPARE HEADS!
• We couldn't find another 1103 spare head, so we opted for the 1125 along with its corresponding controller. This worked fine, except the unit is thicker and the whole beam path changed by ~ 1 cm in height. Mayank and I restored the alignment and centered the ITMX oplev. A photo of the changes made to reach the final setup are in Attachments #1-2. Most notably we had to lower the big pedestals where the HeNe laser head is resting to accommodate this height difference.
• Finally the laser beam diameter is different, so we removed the "collimating" lens immediately after the HeNe head. This had the effect of restoring the beam path, but we placed a shorter focal length lens between the last steering mirror and the QPD> this means the previous ITMX oplev calibration is now bogus.

{Mayank, Yehonathan}

Yesterday, we measured AS55 and REFL11 dark noises at the IQ demod boards outputs (attachment 1) using SR560+SR785 setup.

We also measured the whitening board noise of REFL11 using an improvised adapter (picture will come later). The measurement result is shown in attachment 2. Didn't have time to measure the whitening noise for AS55

Also, after realizing the Finesse model doesn't account for the REFL port attenuation I measure how much DC power at the REFL11 PD to be 0.8mW by aligning PRM and misaligning the rest of the optics.

For some reason, the power before the attenuation is only ~ 360mW. The Finesse model predicts around 700mW. Where is the rest of the light going?

I added the PDs Dark noises (using the recently measured IQ demod gains) and shot noise to the PRMI carrier noise budget (attachment 3). ADC and whitening noises coming soon.

Measurements and PRMI noise budget notebooks were uploaded to the 40m git.

{Paco, Yehonathan, Yuta}

Paco and Yuta locked PRMI carrier and I took the MICH OLTF measurement (attachment 1).

I downloaded 300secs of C1:LSC-MICH_IN1_DQ from when the PRMI was locked yesterday and calibrated it with the OLTF. I plot it together with the noise budget (attachment 2).

{Mayank, Yehonathan}

We measured the noise at the WF1 (REFL11) and WF2 (AS55) boards at the LSC rack with and without whitening filter. We switch the filter on and off by switching off and on the unwhitening in the PDs filter bank.

Attachment 1 shows the measurements.

Attachment 2 shows the ratio between the noise with and without whitening filter. I also plot the inverse of the unwhitening MEDM filter (all the unWhite filters were the same). I tune the gain of that filter to match the ratio of the AS55 whitening noises.

This is because I couldn't match the ratio of the REFL11 noises.

Moreover, the overall gain doesn't make sense to me. AS55 whitening has a gain of 24db and REFL11 has a gain of 18db. I'm not entirely sure where these values should show up. Also seems like REFL11 whitening has more gain than AS55 whitening. Will have to investigate more tomorrow.

{Mayank, Yehonathan}

We measured today the TFs of the whitening boards. We measured in particular REFL11 I/Q and AS55 I/Q channels using SR785.

There seems to be an issue with turning on whitening gain bigger than 18dB. In all our measurements, when the whitening filter was off the TF was flat and had the right gain. However, when we turned the whitening on, the measured TFs for gains higher than 18 dB would like exactly like as if the whitening gain was 18 dB. This happened in all channels that were measured and across two separate whitening filter boards.

Also, it was hard to measure both low and high-frequency parts of the TFs when the gain was high. The gain difference should be normally 40 dB but for higher gains it seems smaller. We verified that at higher gain level the high-frequency response was dependant on the ecxitation level meaning we had some saturation there.

We forgot to take a reference TF measurement by looping the SR785 on itself using the same BNC cables used for the actual measurement. I took this measurement today (attachment 1). As can be seen, there is a significant delay in the SR785 + cables themselves.

I also retook some measurements on the AS5_I whitening channel for various gains. Being careful with the excitation level and the channel range on the SR785 to avoid saturation I was also able to see low-frequency gains higher than 18dB so that problem is gone too. The results are shown in attachment 2 with the reference phase subtracted from the measurements.

[Mayank, Radhika]

I retook a transfer function measurement of the uPDH servo closed-loop (using the SR560 to simulate a cavity pole) [Attachment 1]. While some coherence is lost at low frequencies, the servo does not appear to be saturating. Moving forward this measurement is used to design a digital filter that can replicate the uPDH servo box response. *Note: for now the chosen sampling frequency for the discrete filters is 61.04 kHz, the lowest sampling frequency setting of the Moku:Go.

We performed a low-order fit of the TF using vectfit. Vectfit always seems to return 1 more zero than pole - this results in an "improper" transfer function that causes any transformation to the z-domain to fail. Mayank took the fitted zeros and poles from vectfit and manually removed one of the zeros. After transforming the zeros and poles to the z-domain (using control.matlab.c2d), we noticed multiple resonances around 100 kHz that reached 10-20 dB. We decided to estimate poles and zeros by eye instead of using vectfit. 

2 zeros and 2 poles were selected by eye to get an estimated fit in the s-domain. Using continous-to-discrete transforms (tried scipy.signal.bilinear and control.matlab.c2d) resulted in unstable controller responses. Attachment 2 shows the original TF measurement with the designed analog filter and the resulting digital filter. The orange 'x's and 'o's mark the poles and zeros used. The digital filter contains many high-frequencies resonances, the most significant at the sampling frequency, 61.04 kHz, reaching 20 dB. Next we tried to manually load the analog ZPK coefficients into Foton. This resulted in the same digital filter as the python s-domain to z-domain functions [Attachment 3].

**UPDATE** Now looking back it's clear that the high-frequency response is limited by the sampling rate. I will redo this for the highest Moku:Go sampling rate of 3.9 MHz.

XAUX laser locked with Moku:Go controller

The analog zeros and poles used to design this filter were:

Attachment 1 shows the resulting digital SOS filter (sampling rate: 3.9 MHz) compared to the measured uPDH servo transfer function (loop closed). The filter design was loaded on the Moku:Go.

Lock acquisition

I locked the AUX laser with the uPDH servo box and maximized its transmission to ~0.8. I then fed the Moku digital filter output to the PZT and the laser was able to catch lock. However, the max green transmission I could achieve using the Moku controller was 0.5. Attachment 2 is a screenshot of the green transmission ndscope during a lock sequence.

I measured the OLTF of the loop by injecting an excitation at the error point. An SR560 was used to sum the error signal with the excitation. The Moku multi-instrument mode was configured with the Frequency Response Analyzer and Digital Filter Box; it was able to source the excitation and take a transfer function measurement of error signal / (error signal + excitation), while keeping the loop closed.

The OLTF measurement [Attachment 3] points to a loop UGF of ~4 kHz, and phase margin of ~70 deg. An optimal controller would be able to boost the gain around the UGF without changing the phase too much (lag compensator)?

I added input noises and angle to length coupling to the noise budget.

I added ADC and whitening filters noise contributions. The ADC noise is assumed to be 1uV/sqrtHz and the whitening noises were measured before in elog 17582. I use the measured whitening filter (elog   17584 ) to get the signal referred noise and calibrate.

The angle-to-length coupling is computed by taking the suppressed OpLev noise spectra of ITMX, ITMY, and BS and converting them to length noise by using the recently measured coupling coefficients in ELOG   17583 

I began commissioning the AS WFS for (PR)FPMI configurations

The main goal of this work is to close some ASC feedback loops during FPMI, or PRFPMI configurations to make locks more robust and their acquisition reliable. The running hypothesis is that angle to length coupling of the differential arms is increasing the rms and therefore straining the LSC controls... Today I began working on this, and unfortunately didn't make a lot of progress due to hardware hurdles, but the nominal plan is:

• Stop IMC WFS and hijack the RFQPD heads for sensing -- WORK IN PROGRESS
  • We can temporarily break the IMC WFS loops to work on this upgrade, but in the future we would like to have at least one head (WFS1 or WFS2?) + transmission QPD to control IMC while the other head is our AS WFS.
  • Anchal already commissioned the AS WFS electronics and models. He also installed a couple of flipper mirrors at the Vertex table that pick the whole AS beam (otherwise going into the AS camera and ASDC, AS55, and AS110 PDs).
    • While this allows single ARM cavity ASC, we can't control MICH using our AS55 error signals.
    • I found that a beam was picked off along the AS110 path for another WFS head (currently unused), so after moving the head out of the way, I installed a mirror that sends it all the way to the IMC WFS heads area. The configuration before and after is summarized in Attachments #1-3.
    • The beam will need profiling again, but this doesn't destroy any of Anchal's ARM ASC alignment, just blocks it so we can recover it, so I think it's a good solution.
• Lock FPMI and measure sensing matrix -- PENDING
• Close loop, aim for gain of 10 at 10 Hz, then roll-off (e.g. IMC WFS) -- PENDING (RXA: G = 10 @ 1 Hz. G = 10 @ 10 Hz would be too spicy)

DESCRIPTION OF ATTACHMENTS

Attachment #1 before the changes were done

• The red beam path indicates the nominal AS beam path
• The purple beam path indicates the AS110 beam path.
• The blue beam path indicates the unused, AS WFS (?) beam path.

Attachment #2-3 after the changes were done

• The blue beam indicates the newly picked AS WFS beam path.
• The yellow beam indicates the ARM AS WFS alternative path (installed by Anchal, controlled by flipper mirrors).

Is the A2L coming from the optical lever feedback? If so, we can make a 30 Hz ELP to cut it off by 60 Hz.

As a test, I have replaced the PSL StripTool on the north wall of the control room with a nearly equivalent NDScope display. It is plotting the real time second trend along with min/max values. Let's try this out for awhile, and if its bad we can just close the window.

FYI, the striptool machine, zita, has a mis-configured network setup so that its not getting a useful nameserver. So for now, its running on allegra through ssh -Y. We should put the zita network thing on some sort of issue tracker. Maybe we can have tickets and issues like various companies use for tech support? This could cover all of our main lab work and help us keep track of little problems like this.

We aligned the AS WFS and measured some angular motion from ETMX

This work is done after IMC WFS offsets are offloaded and the WFS loops are turned off. Then, the flipper mirror M4 in the IMC WFS path is flipped up to feed the AS beam into its path. We also connected the AS WFS demod boards for both heads.

[JC, Paco]

We continued to get the AS WFS going. We added two lenses with the help of JC (f1~500 mm and f2~1000 mm) to have control over the minimum waist and its position downstream. In practice, we placed the lenses such that the beam waist lands close to the last lens before WFS1, giving a reasonable beamspot at both WFS1 WFS2 heads. These changes are reflected by the picture on Attachment #1. The beam sizes will sure need careful optimization to sense the Differential/Common ARM angular motion due to ITMs and ETMs, but for now we decided to move forward.

[Mayank, Paco]

We locked FPMI and use the WFS LOCKINs to inject a 21.173 Hz line into ETMX-PIT. Then, we look at the AS11-WFS signals using diaggui to identify any measurable signal. We see the line coming into a combination of channels C1:ASC-DHARD_PIT_IN1, C1:ASC-DHARD_YAW_IN1, C1:ASC-DSOFT_PIT_IN1, and C1:ASC-DSOFT_YAW_IN1 as shown by Attachment #2.

It's time to start commissioning the new coil drivers. The Acromag box is already there but it needs to be modified.

The idea is to take the channel configuration from C1AUXEY that controls a single suspension - ETMY and apply it for each suspension in 1X4, that is PRM, BS, ITMX, ITMY.

For this, we need the following items

Currently, there are 8 spare BIO channels in the existing Acromag, we will need 12 more which requires a new BIO Acromag.

Wiring feedthrough spreadsheet coming soon.

The fan behind the PSL controller is injecting excess band limited noise

Background

While doing noise hunting to improve the BHD lock stability, we noticed peculiar noise bumps in the BH44 error point near (but not exactly at) the even line-harmonics (for example 120 Hz, 240 Hz, ...). Other channels such as C1:HPC-BHDC_SUM, C1:LSC-PRCL_IN1, or even C1:LSC-XARM_IN1 didn't show these features, so we looked at C1:IOO-MC_F_DQ (which represents the free running laser noise above 100 Hz) and to our surprise found this excess noise!

Noise hunting around PSL

Since this noise is present upstream, we decided to hunt around using C1:IOO-MC_F_DQ. We set up a diaggui measurement to do some "live demodulation" as suggested by Koji in order to understand the nature of this noise. In order to get some "video bandwidth" we set up a power spectrum measurement from 114 Hz to 140 Hz (to monitor the usual 120 Hz line noise peak) with a bandwidth of 1 Hz. A single exponential average gave us the 1 Hz narrow spectrum in "real time", from which we noticed its nonstationary character. The band limited excess noise is the result of a peak hovering in the range of 125 to 131 Hz. With this diagnostic set up, we started hunting for its source.

1. We checked the fan behind the PSL controller (Attachment #1). 
   1. After disconnecting the molex powering it with +15 VDC the noise dissappeared!

To show the impact in the complete noise spectrum, we took a 10 fixed average measurement with and without the fan being on. The result is in Attachement #2. The spectra are shown along with their rms, which is significantly reduced when the fan is off near the 100 Hz frequency band (where these bumps appear). Anyways, we have left the fan on because the PSL controller needs it so the problem remains, but we have at least identified the source.

This is super! And now is the time to replace the internal fan!

We removed the PSL controller internal (broken) fans after it tripped due to overheat.

Background

[Mayank, Radhika]

While aligning Xarm I noticed sudden loss of beam. On Radhikas suggestion we cheked the PSL and found out that the PSL controller was in off state (No lights on front and back panel). We restored the situation by unplugging and replugging the power cord. The PSL worked fine for a few minutes (~ 30 ) and then tripped again.This time the front panel  OFF light was on . See attached image (Attachment #1).

Fix

[Paco, Mayank, Koji-remote]

We disconnected the PSL controller and took this opportunity to investigate the controller's internal cooling mechanism. After disassembling the top panel of the chassis, we saw there are two SUNON - KD1205PHB2 fans meant to run at 12 VDC (1.7 W) connected to the bottom pcb inside the controller. After disconnecting them from this board, we tested them with an externally supplied dc voltage and confirmed they no longer worked. We noted the cooling mechanism is based on a long aluminum heat sink to which most ICs are attached, and the fans are meant to provide airflow towards the rear aperture on the chassis. We followed Koji's suggestion and for now, removed the damaged components (detailed pictures of this operation have been posted in a google photos album elsewhere) to allow heat to flow out more easily.  We reassembled the controller chassis and reinstalled it with the external fan providing the necessary airflow to prevent the unit from tripping again due to overheating. Then we turned on PSL and recovered PMC and IMC locks.

Aftermath

We took a C1:IOO-MC_F_DQ trace after this work to confirm our earlier findings; the trace is attached in Attachment #2. The noise bumps are present as expected. This is still not a desirable configuration so next step would be replacing the external fan, or even better, find the appropriate spare of the internal units and berid the external one.

We lowered the BH44 and BH55 DC transimpedances to ~ 50 V/A

[Paco, Yuta]

Background

When locking the homodyne phase angle using BH44_Q or BH55_Q error signals, we notice the orthogonal quadrature (BH44_I, BH55_I) sometimes appears too noisy. The origin of this useless signal is not known, but we have recently attenuated these beams by placing ND filters before the two RFPDs to avoid saturation effects which become obvious when we lock PRMI. We decided to investigate further by the following tests:

• Remove ND filters and lower the DC transimpedances to ~ 50 V/A
• Check for scattering from suspended optics, e.g. by injecting a line at ITM PIT/YAW and look at the BH44/BH55 demodulated spectra
• Check for PRCL sensing by BH44/BH55, e.g. by measuring the transfer function and/or running a simulation in finesse.
• Check the RF spectra for the signals entering the IQ demod boards (including the 44 and 55 MHz LOs).

DC transimpedance modifications

The first thing we did was change the DC transimpedances of both RFPDs. After removing them from the table, we checked the schematics for 40m RFPDs on the wiki. The DC transimpedance for these gold RFPDs (D980454-v1-C) is estimated as (R22*(1+R13/R23), where these resistors are located around the follower and non-inverting amplifier stages along the DC output traces. After opening the two RFPDs and taking photos of the circuits before any changes (Attachment #1-2), we estimated the DC transimpedances from the measured values for R22, R23 and R13 and summarized them below:

The changes were made on R13 (photos in Attachments #2-3) and the final values summarized below:

All changes have been summarized and recorded in the wiki. The ND filters were set to 0.04 (minimum attenuation) and RFPDs reinstalled.

Next steps:

Continue investigating these items:

• Remove ND filters and lower the DC transimpedances to ~ 50 V/A
• Check for scattering from suspended optics, e.g. by injecting a line at ITM PIT/YAW and look at the BH44/BH55 demodulated spectra
• Check for PRCL sensing by BH44/BH55, e.g. by measuring the transfer function and/or running a simulation in finesse.
• Check the RF spectra for the signals entering the IQ demod boards (including the 44 and 55 MHz LOs).

BH44 is sensitive to PRC alignment noise

[Paco, Yuta]

We investigated the content of BH44 demodulated signals under PRMI configuration. We had a few ideas of what was being sensed by BH44_I but we wanted to test this. Attachment #1 shows a timeseries screenshot of the DCPDs and BH44 error signals during PRMI lock stretch. It is pretty clear how BH44_I is sensing the same as REFLDC. To understand what REFLDC is sensitive to, we locked PRY (this is like having a lossy PRC) and looked at REFLDC, and BH44 error signals again. When PRY is aligned nicely, BH44 error signals show clean LO fringes and we could lock LO_PHASE stably (Attachment #2). Dithering the PRM YAW at 0.5 Hz (amplitude of 150 counts) is sensed by the REFLDC output, so we can attribute its fluctuations to the PRC misalignment (Attachment #3). Now we saw that the zero crossing of the homodyne phase angle changes following REFLDC, and LO_PHASE could not be locked stably. These suggest that alignment of PRC is sensed by BH44, and we might need alignment control to stably lock LO_PHASE in PRMI.
To get the idea of what is causing alignment fluctuations of PRC, we checked the spectrum of SUSPIT/YAW of PRM, PR2, PR3, BS, ITMX, and ITMY. It was not clear what is causing REFLDC fluctuations. (But we found that ITMX and ITMY has huge bounce mode at 16.2 Hz; see Attachment #4).

Next:
 - Check FINESSE to see what BH44 sees. PRCL? PRG?
 - Commission REFL WFS for alignment control of PRC?
 - Commission dither loops (add option to demodulate PRCL, modulate PR2 and PR3) for alignment control of PRC?
 - Check RF spectrum of BH44 and RF LO for 44 MHz (sidebands other than 44 MHz might be contaminating the signal).
 - Check ITMY scattering. Dither ITMY in YAW and check BH44.
 - Move on to PRMI sideband BHD

The PSL wavelength is 1064.5068 +- 0.0015 nm

[Paco, Yehonathan]

With yehonathan's help, we borrowed a WS6-200 wavemeter and calibrated the PSL wavelength using the frequency doubled beam at the PSL table. For this we gathered a FC/APC to FC/PC fiber, cleaned both ends, and launched ~ 6 uW of power into the wavemeter (Attachment #1 shows the setup). The vacuum wavelength was measured to be 532.2534 nm, (see picture in Attachment #2) implying a PSL wavelength of 1064.5068 nm. This is not too far from my previous estimate! The wavemeter claims a "200 MHz accuracy" so we used this as a standard error to estimate the uncertainty of  2 * 0.7 pm @ 532 nm = 1.5 pm @ 1064 nm. This leaves us with 1.4 ppm of relative wavelength error in the ALS based calibration.

• It should be noted that the Nd:YAG temperature was set to 30.615(5) degrees (?) and the injection current to 2.100 Amps for this measurement.

We Changed The Fans on the PSL

To start, the part Koji ordered is 259-1818-ND from DigiKey. This is a Maglev fan from SUNON that should give us less noise. We have 3 spare replacement fans in case these go bad which are stored in the ___ Cabinet along the Y arm (This will be updated once I find a suitable storage spot for the part.)

Starting the replacement process.

Removing the PSL 

    1. Our first step to doing this was to prepare for removing PSL. We began by doing a 60s MC WFS Relief. This will allows us to turn off WFS and close the PSL Shutter next. This is to prevent a large kick once we place the PSL back in its place. 
    2. Went to the PSL and mushed the off button and turned the key. After this, begin by removing the external fan which is shown in elog 17595. After, continue by unplugging the cables from the back beginning with the power cable. Attachment #1 shows the original positions of the cables connected before removing any. Keep in mind, DO NOT TO TOUCH THE KNOBS. If the inputs are changed, this will throw off the beatnotes of the AUX lasers.
        a. There is a black plug at the bottom with a screw hat is hard to reach. Be very patient taking this off because the position of the cable blocks a screqwdriver from untightening the screw. 
        b. A second person should be on the other side of the table to push the module back into arm's reach. Also to make sure the module does not slide back and fall. 
    3. After removing the module, bring into the control room and place onnto the workbench. Make sure all of the red lights are off and the PSL table is closed properly.

Changing the Fan 
    
    
    1. We removed the top cover of the module and opened it all up. Similar to what is shown in elog 17452.
        a. Keep in mind to wear a grounding wristband when working on this.
    2. After removing the old fans and attempting to install the new ones, the holes did not line up correctly (Shown in Attachment #2). To accomodate for this, we used 6-32 screws which gave us just enough slack to fit in all 4 corners.
    3. Ones the fans were bolted down onto the aluminum plate, I soldered the cables to connecting the fan cable to the cables those of the original PSL fans.
    4. Next I used heat shrinks to cover the bare soldered areas and placed the fans into the module. 
    5. We tested the fans by plugging intp the PSL and turning the key. The fans turned on nicely and we proceeded to put it module back together.

Placing the PSL back in its place. 

    1. We place the PSL back into its original spot and began to connect the cables. Make sure the Power cables is put in LAST. 
    2. After the module was put back, we DID NOT put the external fan back into its place. This is to see if the fans which were installed are good enough to maintain the PSL.
    3. Turn the key and press the on Button.

The noise from the external fan is no longer appearing as shown in attachment #3. The PSL has been on for ~2 hrs now and has not turned off. It seems that the fans are doing their jobs well. 

c1sus2 crashed again. Following 40m/17335, I fixed it by running

controls@c1sus2:~$ rtcds restart --all

"global diag reset" made all FE STATUS green burt restored at 2023/May/28/00:19 for c1sus2 models, and watchdogs reset for BHD optics and now all look fine.

Other optics, including MC1, MC2, and MC3 were not damped, so maybe c1sus crashed too I also ran

controls@c1sus:~$ rtcds restart --all

"global diag reset" made all FE STATUS green, burt restored at 2023/May/28/00:19 for c1sus models and watchdogs reset for all other suspensions.

TT1 and TT2 were not responding, and the DAC monitors were frozen... so I ran

./opt/rtcds/caltech/c1/Git/40m/scripts/cds/restartAllModels.sh

I calculated the sensing matrix for PRMI carrier using the Finesse model (git updated) using MAXTEM=2. PRG is calculated to be 11.14, consistent with observations.

The RFPDs were assumed to have a demodulation angle that maximizes the signal they are intended to sens (using MAXTEM=2).

LO Phase is chosen such that it maximizes MICH signal on BHD_DIFF.

That is,

BH44/55 maximized for HPC

REFL11 maximized for PRCL

AS55 maximized for MICH

Values are in uW/nm

50.00

73.67

0.68

17.28

440.80

3.81e-14

0.38

4.22

0.4

3.17

13.46

Some interesting numbers here. First, BHD_SUM is sensitive to MICH. It's not surprising because PRM reflects the MICH signal into POP.

Also interesting, BHD_SUM is super sensitivef to PRCL. Much more than REFL11. We can use it to enhance the PRCL lock.

Unfortunately, although BH44/55 are sensitive to HPC (LO phase), they are swamped by MICH and PRCL. This issue needs to be addressed in order to gain robust LO Phase locking in PRMI.