I have hooked the ALS beatbox into the c1ioo DAQ.  In the process, I did some rewiring so that the channel mapping corresponds to what is in the c1gcv model.

The Y-arm beat PD is going through the old proto-DFD setup.  The non-existant X-arm beat PD will use the beatbox alpha.

Y coarse I (proto-DFD) --> c1ioo ADC1 14 --> C1:ALS_BEATY_COARSE_I
Y fine   I (proto-DFD) --> c1ioo ADC1 15 --> C1:ALS_BEATY_FINE_I
X coarse I (bbox alpha)--> c1ioo ADC1 02 --> C1:ALS_BEATX_COARSE_I
X fine   I (bbox alpha)--> c1ioo ADC1 03 --> C1:ALS_BEATX_FINE_I

This remapping required coping some filters into the BEATY_{COARSE,FINE} filter bank.  I think I got it all copied over correctly, but I might have messed something up.  BE AWARE.

We still need to run a proper cable from the X-arm beat PD to the beatbox.

I still need to do a full noise/response characterization of the beatbox (hopefully this weekend).

I found that the LSCoffset script didn't work today. The script is supposed to null the electrical offsets in all the LSC channels.

I went through the sentences in the script and eventually found that the tdsavg command returns 0 every time.

I thought this was related to the test points, so I ran the following commands to flush all the test point running and the issue was solved.

EDIT, JCD 11June2012:  3rd line there should just be [diag]> tp clear *

Somehow the angular stability of the central part have not been so great.

Also the angular motions look fluctuating a lot and they seem to be related with the glitches.

I took the oplev spectra when the PRMI is locked and unlocked to see whether if something obviously crazy is going on or not.

They seem ok to me except that the PRM pitch shows an extra bump at around 2-3 Hz when the PRMI is locked. But I don't think it's prominent.

- The attached files show the oplev spectra. When the PRMI is locked the PRM and both ITMs are under the length control.

(red) pitch when PRMI is locked

(blue) yaw when PRMI is locked

(orange) pitch without any length controls

(cyan) pitch without any length controls

Last night I took a closer look at the LSC analog signals to find which components are making the glitches.

I monitored the RFPD output signals and the demodulated signals at the same time with an oscilloscope when the PRMI was kept locked.

Indeed the RFPD outputs have some corresponding fast signals although I only looked at the RELL11 I and Q signals.

(REFL33 didn't have sufficiently a high SNR to see the glitches with the oscilloscope.)

I will check the rest of channels.

 Batteries replaced and cylinder recharged. Please clean up your experimental set up if it is blocking breakers or entry way etc.

I will start the final clean up 2pm today.

 The BS and the PRM have 3.3 Hz resonant gain filters that kill the phase margins.

Xarm is aligned for both IR and green. 

Here is a photo of the beam paths of the PSL beat setup.  I want to make sure that the X-green BBPD sees a nice beam from both the PSL and the Xarm, without disturbing the currently working Y setup.  I keep getting confused with all the beamsplitters, especially the green PBSes, which operate at ~56deg, not 45deg, so I made a diagram.

It appears that the old PSL fast channels never made it into the new DAQ system.  We need to figure out what to do with them.

A D990155 DAQ Interface card in far right of the 1X1 PSL EuroCard ("VME") crate is supposed output various PMC/FSS/ISS fast channels, which would then connect to the 1U "lemo breakout" ADC interface chassis.  Some connections are made from the DAQ interface card to the lemo breakout, but they are not used in any RTS model, so they're not being recorded anywhere.

An old elog entry from Rana listing the various PSL DAQ channels should be used as reference, to figure out which channels are coming out, and which we should be recording.

The new ALS channels will need some of these DAQ channels, so we need to figure out which ones we're going to use, and clear out the rest.

Meh.  I've searched in steps of 20 counts in C1:GCX-SLOW_SERVO2_OFFSET units (16 bit +\- 10V DAC, and 1GHz/V coeffecient for the Xgreen aux laser means this is ~0.6MHz per 20 count step).  I went from -400cts to +800 cts and haven't found the beatnote yet.  Meh. 

Both PSL green and Xgreen beams are going to the Xgreen BBPD.  Both beams are easily visible, so while I didn't actually measure the power, it should be sufficient.  The arm is being re-locked in green for each step, but it's not locked in IR, but that doesn't matter for just finding the beatnote.

I've got the output of the BBPD directly connected to the 50 ohm input of the HP8591E spectrum analyzer, with the freq span from 10MHz to 120MHz.  The BBPD is supposed to be good up to ~100MHz, so I should catch any beatnote that's there.  I have to head out, so I guess I'll continue the search tomorrow. 

One of Kiwamu's suggestions was that, since no one is using the Ygreen concurrent with my fiddling, I rotate the waveplate after the PSL doubling oven so that max power goes to the Xgreen path, thus giving myself a bigger signal.  I'll try that tomorrow.  Today, I didn't ever touch the waveplate.

For some reason there appears to have been a spontaneous timing glitch in the c1lsc IO chassis that caused all models running on c1lsc to loose timing sync with the framebuilder.  All the models were reporting "0x4000" ("Timing mismatch between DAQ and FE application") in the DAQ status indicator.  Looking in the front end logs and dmesg on the c1lsc front end machine I could see no obvious indication why this would have happened.  The timing seemed to be hooked up fine, and the indicator lights on the various timing cards were nominal.

I restarted all the models on c1lsc, including and most importantly the c1x04 IOP, and things came back fine.  Below is the restart procedure I used.  Note I killed all the control models first, since the IOP can't be restarted if they're still running.  I then restarted the IOP, followed by all the other control models.

controls@c1lsc ~ 0$ for m in lsc ass oaf; do /opt/rtcds/caltech/c1/scripts/killc1${m}; done
controls@c1lsc ~ 0$ /opt/rtcds/caltech/c1/scripts/startc1x04
c1x04epics C1 IOC Server started
 * Stopping IOP awgtpman ...                                                                      [ ok ]
controls@c1lsc ~ 0$ for m in lsc ass oaf; do /opt/rtcds/caltech/c1/scripts/startc1${m}; done
c1lscepics: no process found
ERROR: Module c1lscfe does not exist in /proc/modules
c1lscepics C1 IOC Server started
 * WARNING:  awgtpman_c1lsc has not yet been started.
c1assepics: no process found
ERROR: Module c1assfe does not exist in /proc/modules
c1assepics C1 IOC Server started
 * WARNING:  awgtpman_c1ass has not yet been started.
c1oafepics: no process found
ERROR: Module c1oaffe does not exist in /proc/modules
c1oafepics C1 IOC Server started
 * WARNING:  awgtpman_c1oaf has not yet been started.
controls@c1lsc ~ 0$ 

I did liso simulation of the circuit in the seis box. I think that AD620 (first amplifier in the circuit) noise might be much less with the signal from guralps from 0.01 Hz. Here is the TF of AD620 output / circuit input.

The noise spectrum is at this node is

The psd of the seismic noise below 1 Hz ~ 1u m/s => circuit input signal is ~1mv.

The TF of the whole circuit is

This result differs from the graph on the circuit sheet, but may be it was done before the resistor parameteres changed. Back of the envelop calculations also show that it is not possible to acheive DC gain 200 while 50-800 Hz gain = 5000. I'll check with the spectrum analyzer.

AD620 might be a weak point in the simulation since this is not a "classical" operational amplifier, it contains a resistor that adjusts the gain. During the liso simulation I assumed that we have an ordinary opamp (with noise, gain and gbw parameters taken from the real ad620 datasheet) with a resistor parallel to the opamp = 50k and a resistor before the inverted input that corrsponds to R2. In this case the gain of the simulated opamp is the same as of the real one given by the formula 1 + 49.9k / R2, though noise parameters may change. This should be also checked with the spectrum analyzer.

Correction list by visiting safety committee, Haick Issaian is not shown:

1,  update laser, crane operator list and post it

2,  check fire extinguishers monthly, date and initials must be on the tags

3,  move drinking water tower so it does not block fire extinguisher

4,  post updated crane doc at cranes

5,  post present phone lists at IFO room phones

6,  emergency laser shutoff at the south end must be mounted without C-clamp

7,  use heavy cable tie to insure position of  mag-fan on cabinet top

Additional to do list:

a,  safety glasses to be cleaned

b, let the electrical shop fix Rack-AC power to optical tables at the ends

c, measure transmission of  laser safety glasses

d, update  IFO outside door info signs

e, update laser inventory and post it

f,  schedule annual crane inspection and renew maintenance contract

g, PSL enclosure  inner shelf needs a good clean up so it is earthquake safe

I found that none of the filter banks in the LSC input signals have the precise anti-whitening filters.

I installed the precise filters on REFL11, REFL33, REFL55 and AS55 based on Jenne's measurement (#4955)

After installing them I briefly checked the REFL11 sensing matrix with the PRMI locked, but it didn't change so much from what I got (#6283).

But I felt that the PRMI became more robust after that ... I just felt so ...

(Background)

(What I did)

(Next ?)

• Check  the amount of of the sidebands using the OSA
• Check the amount of the DC light
• Check the sensing matrix to see if the absolute values in watt / meter make sense or not
  • This work needs calibrations on all the demodulated board (this is equivalent to measuring the conversion losses of the mixers in the demod boards).
• Measure the contribution from the RAMs (it must be measurable by some means)

I installed the rest of the precise anti-whitening filters. Now all of the LSC sensors have the right filters.

 Just to be clear what I said at the meeting, I write all this down here. Adaptive filtering of real signals (MC_F and GUR1_X) with all noises inside is 

This is offline filtering but with real signals and with the C-code that is compiled at the 40m now. We can reduce the MC_F signal by ~100 below 10 Hz, but  the problem is that reducing the adaptation gain, the error increases. As a result when we move towards FxLMS algorithm with AA, AI and downsampling, we have to take the gain equal to ~1e-2 and we do not reduce any noise. 

The second demonstration of this problem is static Wiener filtering. This is the result

We can see that adaptive filtering outperforms the "optimal" filtering. This is because an adaptive filter can follow the changes of coefficients immediately while the Wiener filter averages them. This is the mathematical formulation:

mcl_real = coeff _real* seismic_noise_real + other_noise

mcl_real - the real length of the MC,

coeff_real - real coefficients, that represent the transfer function between seismic noise and MC length,

other_noise - noise uncorrelated to the seismic noise seismic_noise_real

But in the world of our measured signals we have the equation

mcl_measured = coeff * seismic_noise_measured + other_noise

mcl_measured = TF_mcl * mcl_real

seismic_noise_measured = TF_seis * seismic_noise_real

where TF_mcl and TF_seis - transfer functions from the real world to measurements.

It seems to me that TF_mcl or TF_seis are not constants and for that reason the TF between measured seismic noise and mcl is not constant. But it is exactly what an adaptive or Wiener filter tries to define:

coeff(time) = average(coeff(time)) + delta(coeff(time))

The result of applying average(coeff) is the green line in the Figure 2 - error after applying the Wiener filtering.

delta(coeff) - the changing part of the transfer function is caught by the adaptive filtering. The lower the gain, the lower is the capability of adaptive filter to catch these changes. Theoretically. the error after applying adaptive filter can be presented like this:

E(error*error) = E(other_noises*other_noises) + 1/(2-mu)*mu*E(other_noises*other_noises)  + 1/ {mu*(2-mu)} * Tr(Q) * A

where mu = adaptation gain

Q - covariance matrix of delta(coeff)

A - norm of the seismic signal

The first term in this equation is the dispersion of other noises, the second term is the error of the adaptive filter due to non-zero gain, the third term is due to the changes in the transfer function - we can see that it is proportional to 1/mu. This term explains why the error increases while mu decreases.

Now I'm looking for the part in the path of the signals where the transfer function can change. As I mentioned above, this is not a change in the real world, it is the change in the measured signlals. My first guess is the quantization error - we do not have not enough counts. If this is not the case, I'll move to other things of the signal path.

The transfer function and current noise were measured.  The location of the peak shifts with the amount of incident light power (RF or DC).  The TF was measured at an incident 1064nm light power of 0.4 mW which produced a DC output voltage of 14 mV => DC photocurrent of 0.28 mA. 

Many of the effects that Koji noted in the previous characterization are still present.

In addition I observed a shift of the peak towards lower frequencies as the RF power supplied to the AM Laser (Jenne Laser) is increased.  This could create a dependance of the demodulation phase on incident RF power.

The plots are attached below.

I briefly ran a Optickle code to see how the PRC macroscopic length screws up the sensing matrix in the PRMI configuration.

Especially I focused on the optimum demodulation phases for the MICH and PRCL signals to see how well they are separated in different PRC length configuration.

It seems that the demod phases for MICH and PRCL are always nicely separated by approximately 90 degree regardless of how long the PRC macroscopic length is.

If this is true, how can we have such a strange sensing matrix ??

(Simulation results)

 Like I said, this is possible if you fail to set up the OSA to look at the sidebands at BOTH the AS and REFL ports at all times. Don't waste your time - set up an OSA permanently!

I am installing an OSA on the AP table and it's ongoing.

I am leaving some stuff scattered on the AP table and I will resume the work after I come back.

1) The REFL165 has been replaced onto the AS table.

2) When the PD interface cable is attached the PD shows a DC out put of 6mV and does not respond to a flash light.  I changed the PD interface port in the LSC rack by swapping the other end of the cable with an unused (Unidentified PD) interface cable,  The PD is working fine after that.   There could be a problem with some binary switch state on the PD interface where the REFL165 cable was plugged in earlier.

I've measured the input signal to the seismic box from seismometer Guralp 1. The spectrum of the signal in the "input +" (TP 1) is

The spectrum below 1 Hz is ~250 uV/sqrt(Hz). As the input is differential, then the input voltage is 0.5 mV/sqrt(Hz). The spectrum of the "output +" signal (TP 2) is

So the gain at ~ 1Hz is ~20. I've measured the transfer function between the "input +" and "output +" (TP1 and TP2) for all 9 circuits

The channels 1-6 are of new modification and have gain ~20 at the frequencies 0.2 - 100 Hz. Below 0.2 Hz the gain is reduced. 100 Hz - cut off frequency of the low-pass filters. Meanwhile channels 7-9 (old configuration) have much more gain and 10_50 Hz filters work here.

The coherence between  "input +" and "output +" (TP1 and TP2) for 9 circuits is

We can see that channel VERT 3 is very bad. For others coherence is lost below 0.2 Hz. The spectrum analyzer noise measured is ~1000 times less then the signal at these frequencies. I'll pay more attention to this loss of coherence at low frequencies. Something is noisy.

 [Koji, Suresh]

To determine the amount of RF power in the AM laser beam at various RF drive levels I measured the RF power out of the Newfocus 1611 PD while driving the AM laser with a Marconi.  During this measurement the DC output was 2.2V.  With the DC transimpedance of 10^4 and a sensitivity of 0.8 A/W we have carrier power as 0.275 mW (-5.6 dBm).  [Incidentally the measured carrier power with a power meter is about 0.55 mW. Why this discrepancy?]

Estimation of the signal strength at the REFL165 PD:

   From the 40m Sensing Matrix for DRFPMI we see that the signal strength at REFL165 in CARM is about 5x10^4 W/m.  Since we expect about 0.1nm of linear range in CARM length we expect about 0.05 mW of RF power.  If the (DC) carrier power is about 10 mW at the photodiode (18mW is about the max we can have since the max power dissipation is 100 mW in the diode)  then the RF : DC power ratio is 5x10^-3 => -23 dB

As this is lower than the power levels at which the PD transfer function was determined and where we noted the distorsion and shift of the resonance peak, it is likely that these effects may not be seen during the normal operation of the interferometer.

The shift due to the carrier power level (DC) change may still however pose a problem through a changing demodulation phase. 

I placed the OSA (Optical Spectrum Analyzer) on the AP table and this OSA will monitor the REFL beam.

Tomorrow I will do fine alignment of the OSA.

(some notes)

- a new 90% BS in the REFL path for limiting the REFL beam power

- Squeezed the ABSL (ABSolute length Laser) path

- Modification of the AS OSA path

 No new stuff for these computers.  Everything should be installed already.

Found it!

The actual temperature of the Xend laser is 0.02 C higher than anticipated based on the formula in elog 3759.  Both the PSL and the Xend laser are at their nominal diode currents (2.100 A for the PSL, 2.003 A for Xend), so the curves should be used as they are.  The PSL temp (when the slow servo offset is ~0) is 31.71 C.  Using curve 2 from elog3759, the Xend laser should be 37.78, which I found was +10 counts on the Xgreen slow servo offset. 

Right now the Xend laser is at 37.80 C, and the beat is around 30 MHz.  This is +80 counts on the Xgreen slow servo.  +60 counts gave me ~80 MHz.  When (a few minutes ago) the MC unlocked and relocked, it came back to a slightly different place, so the temp of the Xend laser had to go up a few 10's of counts to get the same beat freq.  Right now the PSL slow servo offset is 0.076 V. 

The HP8591E is set with ResBW=100kHz, Ref Level= -39dBm (so I'm not attenuating my input signal!).  The largest peak I see for the beatnote is -66dBm.  The nose floor around the peak is -83dBm.  Trace (trace button!) A is set to MaxHoldA, and Trace B is set to ClearWriteB, so B is giving me the actual current spectrum, while A is remembering the peak value measured, so it's easier to see if I went past the peak, and just didn't see it on the analyzer. 

Also, I went back and realigned the beams earlier, to ensure that there was good overlap both near the BS which combines the PSLgreen and Xgreen beams, and at the PD.  The overlap I had been looking at was okay, but not stellar.  Now it's way better, which made the peak easier to see.  Also, also, the waveplate after the doubling oven on the PSL table is still rotated so that I get max power on the Xgreen side of things, and not much at all on the Ygreen side.  I'll need to rebalance the powers, probably after we make sure we are seeing the beatnote with the BeatBox.

Next Steps:

Lay a cable from the BBPD to the BeatBox in 1X2, make the BeatBox do its thing.

Use the dichroic locking to do a sweep of the Xarm.

I've moved GUR1 seismometer from MC2 to the working tables in order not to disturb the MC while working with the seismometer box. The new place for the GUR1 for a few days is near the printer, cables and blue boxes. I've cleaned all mess and wires from the floor, so that seismometer now looks like that

I've connected 2 inputs of the N/S 1 circuit of the seismometer box with a 50 Ohm resistor and measured the noise at the output. The comparison with the seismic signal is

The noise increased at 0.5 Hz and is pretty big. This might explain the loose of coherence at low frequencies.

 Green welding glass 7" x 9"   shade #14 with 40 mm hole and mounting fixtures are ready to reduce scatter light on SOS

PEEK 450CA shims and U-shaped clips  will keep these plates damped.

 This is because spectrum analyzer did not plot the real noise spectrum at the first few points at low frequencies. I've remeasured the noise at 1mHz - 3Hz at "output -" (TP9) and compared it to the seismometer signal

The noise seems to be much less then the signal. I've measured the noise several times and once I got a huge amount of noise

I made another measurement in some time and got the low noise again. A circuit might have a bad contact somewhere.

The plan is to change AD620 adjustable resistor (R2) from 5.49kOhm to 500Ohm to increase the gain from 20 up to 200.

I've replaced R2 resistor that adjusts the gain of the AD620 amplifier. Previous value 5491Ohm, new value 464Ohm, so the gain should increase up to ~200-250. Only at the N/S 1 circuit!

LISO simulation of the circuit transfer function and noise are

LISO predicts gain ~45-46 dB = 200 and noise at the level of 10uV at 1Hz. The transfer function and noise measured are

The noise measured is 5 times higher then predicted by LISO. Though I described AD620 as an ordinary amplifier with 49.9kOhm resistor connecting output and inverted input. I specified the noise spectrum 10 nV and 1/f corner frequency 30 Hz. In the AD620 datasheet noise spectrum is 10 - 100 nV depending on the gain. However, the gain is 200 and noise spectrum should be 10 nV. May be in reality it is not the case. It also possible that the noise model used by LISO is not valid for AD620 as it is not an ordinary operational amplifier.

Today I've attended the laser safety seminar.

I've put the seismometer box back to the 1x1, Guralp is back under MC2. When the seismometer is not plugged in, the noise is

Now, I'm going to collect some data from GUR 1 and MC_F and see if the problem with adaptive filter (increasing errror while decreasing mu) will be gone.

I realigned the steering mirrors for the PMC. The trans value went up from 0.79 to 0.83.

The misalignment was largely in the pitch direction.

PZT1 started railing in the pitch direction and because of this TRY doesn't go more than 0.7. I will leave it as it is for tonight.

Tomorrow I will shift the alignment of the MC to make the PZT1 happier.

The OSA for the REFL beam is now fully functional.

The only thing we need is a long BNC cable going from the AP table to the control room so that we can monitor the OSA signal with an oscilloscope.

The attached picture shows how they look like on the AP table. Both AS and REFL OSAs are sitting on the corner region.

A long BNC cable was installed and now the REFL OSA signal is happily shown on an oscilloscope in the control room.

I checked the laser powers on the AP table and confirmed that their powers are low enough at all the REFL photo diodes.

When the HWP( which is for attenuating the laser power with a PBS) is at 282.9 deg all of the REFL diodes receives about 5 mW.

This will be the nominal condition. 

If the HWP is rotated to a point in which the maximum laser power goes through, the diodes get about 10 mW, which is still below the power rate of 18 mW (#6339).

I used the Coherent power meter for all the measurements.

The table below summarizes the laser powers on the REFL diodes and the OSA. Also the same values were noted on the attached picture.

[Alex / Den]

I've encountered a problem that C1:PEM-SEIS_GUR1_X_IN1 is saved in the int format. It turned out that inside the code the signal is also in the int format.  It is not just a saving error. It should not be so as ADC works at 64k and the model runs at 2k.

Why? There is a bug somewhere in the generation of the code. c1pem.c looks suspicious to Alex because there is a mismatch in the ADC numbers with the simulink model.

Solution: upgrade to 2.4 version - most probably it was fixed there. If not, Alex will handle this problem.

I have slightly shifted the MC beam pointing to relax the PZT1 PITCH. As a result the TRY value went to 0.97 in a first lock trial.

However another issue arose:

   The polarity for controlling the PZT1 PITCH seems to have flipped for some reason.

Since it is still sort of controllable, I am leaving it as it is.

If I remember correctly, sliding the PZT1 pitch value to the positive side brought the beam spot upward in the AS CCD. But now it moves in the opposite way.

Also the ASS feedback looks tending to push the PZT1 pitch to the wrong direction.

I am not 100 % sure if the polarity really flipped, but this is my current conclusion.

(MC pointing)

1. Locked the Y arm and aligned ITMY and ETMY with the ASS servos such that the beam spot on each test mass is well centered on the test mass.
   • With this process the eigen axis of the Y arm cavity is well prepared.
2. Checked the beam positions of the prompt reflection light and cavity leakage field in the AS CCD.
   • It looked the incident beam needed to go upward in the CCD view.
3. Offloaded the MC WFS feedback values to the MC suspension DC biases in a manual way.
4. Disabled the MC WFS servos. The MC transmitted light didn't become worse, which means the suspensions were well aligned to the input beam
5. Changed the DC bias in the MC2 PITCH, to bring the beam spot upward. I changed the DC bias by ~ 0.1 or in the EPICS counts.
6. Aligned the zig-zag steering mirrors on the PSL table to match the incident beam to the new MC eigen beam axis.
   • The transmitted DC light and reflected DC values went back to 27000 counts and 0.58 counts respectively without the WFS servos.
7. Re-engaged the WFS servos.

 I wrote an RAM simulation script ... it calculates the LSC signal offset and the operation point offset depending on the RAM modulation index.

Configuration : RAM is added on optC1, by the additional Mach-Zehnder ifo before the PRM.

 Both are for PRCL sweep result. Note that REFL33I is always almost zero. Next step: Check the LSC matrix with matrix at the offset operation point.

This a kind of self record...

We need an RF setup at POP to extract 22 and 110 MHz components separately.

I am planning to work on this in the daytime on Tuesday.

AS55Q and AS55I signals. AS55Q is around zero while AS55I has a large offset which is about the signal amplitude. It is likely because of the RAM?

keiko, kiwamu

[Keiko / Kiwamu]

 Update on the MICH characterization:

• The OSAs weren't so great because the 11 MHz sidebands were covered by the carrier's tail
  • It seemed that the frequency resolution depended on the mode matching. We will try improving the mode matching tomorrow.
• The noise budget looked very bad
  • There were huge peaks at 1 Hz, 3 Hz, 16.5 Hz and 23 Hz. Something is crazy in the vertex suspensions.
  • Keiko will post the calibrated noise budget.
• The MICH response at AS55Q was measured and we will calibrate it into watts / meter.

 On the right figure, you see the non-zero operation points even when RAM mod index = 0. Apparently they come from non-zero loss of the model.  (Each mirror of 50ppm loss was assumed).

Has default inmat:

PRM, ITMX

Has fancy inmat:

BS, ITMY, SRM (but side is non-fancy), ETMX, ETMY, MC1, MC2, MC3

So it's likely that the MICH problems (giganto 1Hz peak) Keiko and Kiwamu were seeing last night had to do with ITMX having the non-optimized input matrix.  I'll try to figure out where the data from the last freeswing test is, and put in a fancy diagonalized matrix.

 As par Kiwamu's request, RF filters for POP22 and POP110 were installed. They are not really nice. We need to replace it with more fancy electronics.
More to come later.

 We found that that bounce (16.1 Hz) and roll (23.5 Hz) modes on the ITMX were much higher than on the ITMY. After some checking, it seems that the bandstop filters for the

SUSPOS, SUSPIT, SUSYAW, and SUSSIDE loops are set to the correct frequencies. However, the OLPIT and OLYAW had not been set correctly. I have copied the SUS filters into the OL filterbanks and reloaded all the filter banks. Attached are the comparison of old, bad, OL with the SUS ones.

The same cockamamie situation was there for the BS & ITMY as well. Although we still don't have the roll mode frequencies listed in the mechanical resonances wiki, I have guessed that the ITMY roll frequency is the same as the ITMX, since they have nearly the same bounce frequency. OL filters for the BS & ITMY are now at the right frequency (probably). Keiko is on top of fixing things for the other optics.

I think this whole notching adventure was in Leo's hands several months ago, but WE forgot to point him at the OLs in addition to the SUS. I blame Kiwamu 50% for not supervising him and Koji by 45% for not supervising Kiwamu. The other 5% goes to someone else. You know who you are.

Here are the OSEM spectrum of MICH suspensions (BS, IX, IY). Bounce and Roll modes are shown on 16 and 23 Hz. The filters for them has been checked.

keiko, kiwamu, Rana

The BS SIDE damping gain seemed too low. The gain had been 5 while the rest of the suspensions had gains of 90-500.

I increased the gain and set it to be 80.

I did the "Q of 5" test by kicking the BS SIDE motion to find the right gain value.

However there was a big cross coupling, which was most likely a coupling from the SIDE actuator to the POS motion.

Due to the cross coupling, the Q of 5 test didn't really show a nice ring down time series. I just put a gain of 80 to let the Q value sort of 5.

I think we should diagonalize the out matrices for all the suspensions at some point.

All oplev qpd quadrons were zeroed by offset  in blocked dark condition.

The Xgreen PD now has a cable going over to the beatbox. Once beatbox characterization is done I can re-find the beat, and we can do some stuff with the beatbox.