Video captures when power recycling cavity is locked (videos 1 & 2) and flashing (video 3). Arms stayed misaligned.

1. CH1 and CH2 are loooking at PRM front and back faces. CH3 and CH4 are looking at POP and REFL

 2. CH1 and CH2 are loooking at PRM front and back faces. CH3 and CH4 are looking at the ITMs

3. CH1 and CH2 are loooking at PRM front and back faces. CH3 and CH4 are looking at POP and REFL

I centered optical levers of ITMX,BS,ETMY. I also change the position of optical levers of ITMX, ETMY, ITMY, BS on Friday night(9/21), of ITMX, ETMY, BS on Monday night. Both are around 6:00 ~ 7:00.But centering on Monday was totally wrong, because I centered with not good IFO alignment.

The attachment is the 5 days trend of the opt lev of ITMX. First gap is alignment on Friday and Second gap is the alignment on Monday. Now I centered after  locking the FPMI.

The attachment 2 is the last 6 hours data.  The gap on 9/25 00:00 and 1:30(UTC)  is because the alignment of the cavity and the last gap is because of  centering of the optical lever.

 [Manasa, Masayuki]

[revised at 10/1 pm 5:00]

As we mentioned in previous entry (elog#9171), the phase margin of ALS control was at most 20 degree. We modified the filter of C1ALS_XARM and C1ALS_YARM. The OLTF is in attachment1. Now the phase margins of both arms are more than 35 degree. I modified the FM5 filters of both servo.

FM5 filter is the filter for the phase compensation. It had the one pole at 1000 Hz and one zero at 1Hz. As you can see in attachment2, it start to lose the phase at 50 Hz. But the UGF of our ALS control loop is higher than 100 Hz, so I changed the pole from 1 kHz to 3 kHz in order to get more phase margin at UGF. The new servo have 10dB larger gain than previous filter at higer than 1kHz, but the control loop do nothing in that region, so it's no problem.

We have phase lag between 2 arms. I used same filters for both arms, so I'm wondering where these phase lag came from.

 [Manasa, Masayuki]

We made a new flowchart of ALS autolocker. We added the additional step to find the beat note frequency. We have to find a way to read the PSL temperature. By reading the PSL temperature we can decide the sweep range for the end green laser temperature with the curve which measured in previous measurement (in this entry)

We have three thresholds of error signal. One is the threshold for checking the arms are stabilized or not. It should be some hundreds count. Another threshold is to check that the suspensions are not kicked. This should be some thousands counts (in flow chart, it is 2K counts). The other is to check the optimal servo gain. If the servo gain is too high, the UGF is also too high and we will not have enough gain margin. The error signal start to oscillate at the UGF. We will check this oscillation and find the optimal gain. In flow chart this threshold is 1K counts.

I wrote the scripts for cdsutils.

1. fft

usage: fft.py [-h] [-c N_CYCL] [-a N_AVG] freq channel [channel ...]

Measures the frequency content of one or more NDS channels at the specified frequecy.
    The measurement results are  magnitude, phase, real part imaginary part, and the standard deviation of the real and imaginary parts.

positional arguments:
  freq        measurement frequency
  channel     list of measurement channel

optional arguments:
  -h, --help  show this help message and exit
  -c N_CYCL   define number of cycles. Default is 10
  -a N_AVG    define number of average. Default is 100

2. tf

usage: tf.py [-h] [-c N_CYCL] [-a N_AVG] [--dB] freq input output

Measures the transfer funtion of one NDS channels pair  at the specified frequecy.
    The measurement results are the coherence, magnitude, phase, real part, imaginary part, and the standard deviation of the real and imaginary parts.

positional arguments:
  freq        measurement frequency
  input       list of measurement input channel
  output      list of measurement output channel

optional arguments:
  -h, --help  show this help message and exit
  -c N_CYCL   define number of cycles. Default is 10
  -a N_AVG    define number of average. Default is 100
  --dB        print the amplitude in dB
 

3.offsets

usage: offset.py [-h] [-t SECONDS] filterbank [filterbank ...]

Zero the offset of one or more filterbank. Take average of OUT16 channel, and put (-1 * average / gain) into offset.  After change offsets, it will turn on the offset.
    example usage) offset -d 25 C1:LSC-AS55_Q C1:LSC-AS55_I

positional arguments:
  filterbank  list of filterbank to zero offset

optional arguments:
  -h, --help  show this help message and exit
  -t SECONDS  define the averaging time. default value is 10

I put these scripts in /opt/rtcds/cdsutils/trunk/lib/cdsutils.
I couldn't put them into cds library, but I will left tomorrow to Japan...

So please help me Jamie or Joe !!

By the way,

I modified the LSCoffset script  (script)/LSC/LSCoffsets.py.

usage: LSCoffsets.py [-h] [-d SECONDS] [--PDlist]

 Zero the offsets of LSC PDs. During taking average, we will close the PSL and green shutter. After zeroing, it will turn on the offsets switch.
    example usgae) LSCoffset2.py -d 20    

optional arguments:
  -h, --help  show this help message and exit
  -d SECONDS  define the averaging time. default value is 10
  --PDlist    print PD list for LSC and exit

i made new directory 'offset_backup' for old offset script. I move all old offset script into there.

But unfortunately that cannot use right now, because cds offsets script is not available yet...

Last Friday, Matt made a frequency discriminator circuit on a bread board in order to test the idea and study the noise level. I think it will work for phase lock acquisition of Green locking.

As a result a response of 100kHz/V and a noise level of 2uV/rtHz @ 10Hz are yielded. This corresponds to 0.2Hz/rtHz @ 10Hz.

The motivation of using frequency discriminators is that  it makes a frequency range wider and easier for lock acquisition of PLLs in green locking experiment.

For the other possibility to help phase lock acquisition, Rana suggested to use a commercial discriminator from Miteq.

(principle idea)

The diagram below shows a schematic of the circuit which Matt has built.

Basically an input signal is split into two signals right after the input, then one signal goes through directly to a NAND comparator.

On the other hand another split signal goes through a delay line which composed by some RC filters, then arrive at the NAND comparator with a certain amount of delay.

After going through the NAND comparator, the signal looks like a periodic pulses (see below).

If we put a signal of higher frequency we get more number of pulses after passing through the NAND.

Finally the pulse-signal will be integrated at the low pass filter and converted to a DC signal.

Thus the amplitude of DC signal depends on the number of the pulses per unit time, so that the output DC signal is proportional to the frequency of an input signal.

(result)

By putting a TTL high-low signal, an output of the circuit shows 100kHz/V linear response.

It means we can get DC voltage of 1 V if a signal of 100kHz is injected into the input.

And the noise measurement has been done while injecting a input signal. The noise level of 0.2Hz/rtHz @ 10 Hz was yielded.

Therefore we can lock the green PLL by using an ordinary VCO loop after we roughly guide a beat note by using this kind of discriminator.

I have set up new summary pages for the 40m: http://www.ligo.caltech.edu/~misi/summary/
This website shows plots (time series, spectra, spectrograms, Rayleigh statistics) of relevant channels and is updated with new data every 30 min.

The content and structure of the pages is determined by configuration files stored in nodus:/users/public_html/gwsumm-ini/ . The code looks at all files in that directory matching c1*.ini. You can look at the c1hoft.ini file to see how this works. Besides, a quick guide to the format can be found here http://www.ligo.caltech.edu/~misi/iniguide.pdf

Please look at the pages and edit the config files to make them useful to you. The files are under version control, so don’t worry about breaking anything.

Do let me know if you have any questions (or leave a comment in the pages).

I have created a wiki page with introductory info about the summary page configuration: https://wiki-40m.ligo.caltech.edu/Daily summary help

We can also use that to collect tips for editing the configuration files, etc.

Thanks for the great entry!

In order to make this work for higher frequencies, I would add Hartmut's suggestion of a frequency dividing input stage.  If we divide the input down by 100, the overall range will be about 200MHz, and the noise will be about 20Hz/rtHz.  That might be good enough... but we can hope that the commercial device is lower noise!

Today we tried the Schmitt trigger DFD, and while it works it does not improve the noise performance.  At least part of our problem is coming from the discrete nature of our DFD algorithm, so I would propose that an industrious day job person codes up a new DFD which avoids switching.  We can probably do this by mixing the input signal (after high-passing) with a time-delayed copy of itself... as we do now, but without the comparator.  This has the disadvantage of giving an amplitude dependent output, but since we are working in the digital land we can DIVIDE.  If we mix the signal with itself (without delay) to get a rectified version, and low-pass it a little, we can use this for normalization.  The net result should be something like:

output = LP2[ s(t) * s(t - dt) / LP1[ s(t) * s(t) ]],

where s(t) is the high-passed input and LP is a low-pass filter.  Remember not to divide by zero.

Power Spectral Density plot using PyNDS, comparing 5 fast data channels for ETMX.

**EDIT** Script here:

import nds
import numpy as np
import matplotlib.pyplot as plt
import time
daq=nds.daq('fb', 8088)
channels=daq.recv_channel_list()
e=0
start=int(time.time()-315964819)
rqst=['C1:SUS-ETMX_SENSOR_UR','C1:SUS-ETMX_SENSOR_UL','C1:SUS-ETMX_SENSOR_LL','C1:SUS-ETMX_SENSOR_LR','C1:SUS-ETMX_SENSOR_SIDE']    #Requested Channels
for c in channels:
    if c.name in rqst:
        daq=nds.daq('fb', 8088)
        data=daq.fetch(start-100, start, c.name)
        vars()['psddata'+str(e)], vars()['psdfreq'+str(e)]=plt.psd(data[0],NFFT=16384,Fs=c.rate)
        vars()['label'+str(e)]=c.name
        e+=1
plt.figure(1)
plt.clf()
plt.title('PSD Comparison')
plt.grid(True, which='majorminor')
plt.xlabel(r'Frequency $Hz$')
plt.ylabel(r'Decibels $\frac{dB}{Hz}$')       
for x in np.arange(0,e):
    plt.loglog(psdfreq0, 10*vars()['psddata'+str(x)], label=vars()['label'+str(x)])
plt.legend()
plt.show()

[Yuta, Paco, Murtaza]

[WIP]

Chasing the excess noise in TRX, Yuta suspected the whitening-dewhitening situation for the ETMX.
We compared the OPLEV spectrum for the IFO optics to gauge the situation. The spectrum with dewhitening filters (p = 1Hz, z = 10Hz) (FM3, FM4) 
-Active for all optics (Attachment 1)
-Deactive for ETMX (Atachment 2)

Attachment 1: ETMX >1Hz shows the extra dewhitening filter applied which concludes a misisng analog whitening filter (which it is supposed compensate for)

We compared the X-end with the Y-end rack and found the whitening board for ETMX OPLEVs missing (Pentek Generic Input Board).
We found the board but could not determine the reason for the missing board in previous elogs.

We proceeded to check the board for potential damage. To do this, we we evaluated the transfer functions the filters.

- The board schematic does nott exist on D020432 anymore and is moved to D1500270 (The label on the board says D020432)
- The schematic does not record the modified values for the resistors and capcacitors to place the pole-zero pair at 1, 10Hz
[INSERT NEW VALUES]
- The transfer functions were evaluated using a swept sine measurement from the {input}-> {1st header, 2nd header, tie point} (for example, {J3} -> {J1, J2, T1}) for the first 4 filters (which had existing connections) (Attachemnt 6)
- A good transfer function ( Attachment 3)  is expected with the filter design was obtained on a few ports ✓; others looked garbage (Attachment 4) ❌
- Attachment 5 shows the good/bad outputs

Summary of QPD filter whitening situations:
 - ETMX and ETMY oplevs have whitening (not now for ETMX) of two 10:1 (D020432, which is actually D1500270)
 - TRX and TRY QPDs have whitening of two 40:4 (D1400415, D1400414)
 - ITMX and ITMY oplevs have whitening of two 10:1 (D020432, which is actually D1500270)
 - BS, PRM, SRM oplevs and MC2 TRANS QPD do not have whitening
 - They are always on and compensated with digital anti-whitening filters (not now for ETMX; for now, ETMX digital anti-whitening filters are turned off to have better oplev damping).

 I tried to calculate the frequency of resonance using Rayleigh's method.  approximated the geometry of lead to be that of a perfect cylinder, and  the deformation in the lead by the deflection in a cantilever under  a shear strain.

this rough calculation gives an answer of 170Hz and depends on the dimensions of each lead, number of leads, and mass of the granite. But the flaw pointed out is that this calculation doesnot depend on the dimension of the granite slab, nor on the exact placing of the lead spheres with respect toteh COM of the slab.

I will put up the calculations details later, and also try to do a FEM analysis of the problem.

BTW, latex launched this new thing for writing pdfs. doesnot require any installations.  check  http://docs.latexlab.org

i added my laptop's mac address to teh martian at port 13 today.

This week I attended a whole lot of orientations, lectures, and meetings related to SURF. Done with general and laser safety training.

read Nergis' thesis for, and other material on WFS.

got confused with how the sidebands and shifted carrier frequencies are chosen for the Interferometer, read initial chapters of Regehr's thesis for teh same.

Made a plan for proceeding with the WFS work through discussions with Koji.

Understood the MC cavity and drew a diagram for it and the sensors.

Did Calculations for Electric field amplitudes inside and outside the MC cavity.

Saw the hardware of the WFS and QPD inside, and their routes to computers. Figured out which computer shows up the conditioned data from teh sensors.

Tried calculating the cavity axis for MC using geometry and ray tracing. Too complicated to be done manually.

Read some material (mainly Seigman) for physics of calculating the eigen-axis of the MC cavity with mirrors mis-aligned. Will calculate that using simulations, using the ABCD matrices approach.

Made a simple feedback simulink model yesterday to learn simulink. Made it run/compile. Saw the behaviour thru time signals at different points.

in the night, Made a simulink model of the sensor-mirror thing, with transfer functions for everything as dummy TFs. Compiles, shows signals in time. Remaining part is to put in real/near-real TFs in the model.

Wednesday Morning E-log :

Most of the time through this week, i was working towards making the simulink model work.

It involved learning simulink functions better, and also improving on the knowledge of control theory in general, and control theory of our system.

1. Thusrday : found tfs for the feedback loop. and tried many different filters and gains to stabilize the system (using the transient response of the system). - not through

2. Friday : decided to use error response and nullify the steady state error instead of looking at convergence of output. tried many other filter functions for that.

Rana then showed me his files for WFS.

3. Sunday - played with rana's files, learnt how to club simluink with matlab, and also about how to plot tfs using bode plots in matlab.

4. Monday : Read about state-space models, and also how to linearize in matlab. done with the latter, but the former still needs deeper understanding.

read ray-optics theory to calculate the geometric sensing matrix.

It first requires to calculate the eigen mode of the cavity with tilted mirrors. this eigen mode is needed to be found out using ray-optics transfer matrices for the optics involved  . figured out  matrices for the tilted plane mirrors, and am working on computing the same for MC2.

5. Tuesday : went to Universal Studios , Hollywood :P

6. Wednesday (today) : Writing the report to be submitted to SFP.

Wednesday after the meeting - Started report, learnt mode cleaner locking from Kiwamu and Rana, saw how to move optics on the tables with Rana and kiwamu.

Thursday - Made the report

Tuesday - report.

Today - am trying locking the MC with kiwamu's help to see the WFS signals and also to start characterizing the QPD.

Nancy and Koji:

This is what I and Koji measured after aligning the MC in the afternoon.

MC_Trans 4.595 (avg)

MC_Refl 0.203 (avg)

MC2_trans :

power = 1.34mW

13.5% width : x=6747.8 +- 20.7 um  , y = 6699.4+- 20.7 um

The WFS error signals were recorded in the order

WFS1_PIT

WFS1_YAW

WFS2_PIT

WFS2_YAW

these measurements are made in the linear region, that is the MC is nearly perfectly aligned.

This is  the average and std. dev.of 5 measurements taken of the same signals over 10 secs each. The std. dev are under 10%. And hence, I will be using 10 secs for measurements for the WFS signals after perturbations to the mirrors.

avg =

829.4408
-517.1884
297.4168
-944.7892


std_dev =

9.0506
22.9317
15.4580
8.9827

I perturbed the Pitch and Yaw of the Three mirrors (in order MC1,2,3), using ezcastep and calculated the coefficients that relate these perturbations to the WFS error signals.

The perturbation made is of -0.01 in each dof , and after measuring the WFS error for it, the system is reverted back to the previous point before making the other perturbation.

I was able to calculate the coefficients since I have assumed a linear relationship..

Following are the coefficients calculated using 10 secs measurements

coef_mat =

   1.0e+05 *

                            MC1_P   MC1_Y  MC2_P   MC2_Y    MC3_P   MC3_Y  constant
WFS1_PIT        -0.1262    0.3677   -0.4539   -0.6297   -0.1889   -0.1356   0.013664
WFS1_YAW     -0.0112   -0.7415   -0.1844    2.4509   -0.0023   -0.3531  -0.016199
WFS2_PIT         0.1251    0.4824   -0.2028   -0.6188    0.0099   -0.1490   0.006890
WFS2_YAW      0.0120   -0.7957   -0.1793    0.9962   -0.0493    0.2672 -0.013695

Also, I measured the same thing for 100s, and to my surprize, even the signs of coeficients are different.

coef_mat =

   1.0e+05 *

                           MC1_P   MC1_Y  MC2_P   MC2_Y    MC3_P   MC3_Y   constant
WFS1_PIT       -0.1981    0.3065   -0.6084   -0.9349   -0.4002   -0.3538   0.009796
WFS1_YAW     0.0607   -0.6977    0.0592    2.8753    0.3507    0.0373   -0.008194
WFS2_PIT        0.0690    0.4769   -0.2859   -0.7821   -0.1115   -0.2953  0.004150
WFS2_YAW     0.0580   -0.8153   -0.0937    1.1424    0.0650    0.4203  -0.010629

The reason I can understand is that the measurements were not made at the same time, and hence conditions might have changed.

A thing to note in all these coefficients is that they relate the error signals to the 'perturbation' around a certain point (given below). That point is assumed to lie in the linear region.

MC1_PIT      2.6129
MC1_YAW   -5.1781
MC2_PIT       3.6383
MC2_YAW    -1.2872
MC3_PIT      -1.9393
MC3_YAW    -7.518

I and Koji were trying to lock the mode cleaner for measuring the beam power at MC2 end. That is when we obtained the trans and refl values.

The beam characteristics at the MC2 were measured so that we could now use a dummy beam of similar power to test and characterize the QPD we are about to install at the MC2 end. This QPD wil provide two more signals in pitch and yaw, and hence complete 6 signals for 6 rotatioanl dof of the cavity. (4 are coming from WFS).

Once the QPD is characterised, it can be used to see the spot position at MC2. This is related to the mirror angles.

The width measurements were done using a beam scan. the beam scan was properly adjusted so that the maxima of the intensity of the sopt was at its center.

We also fitted gaussian curve to the beam profile, and it was a substantially good fit.

The whole idea is that I am trying  to look how the Wavefront sensors respond to the perturbations in the mirror angles. Once this is known, we should be able to control the mirror-movements.

the starting point would be to do just the DC measurements (which I did today). For proper analysis, AC measurements are obviously required.(will be done later).

The matrices so calculated can be inverted, and if found enough singular, the method can be used to control.

The first shot is to see teh dependency of teh error signals only on MC1 and MC3, and see if that is kind of enough to control these two mirrors.

If this works, the QPD signals could be used to control MC2 movements.

I just found the singular values and the condition number of the 4*4 matrix relating the WFS error signals and the MC1 and MC2 movements.

the condition number is ~12.5. I think its small enough to continue with the scheme. (if the measurements and all are reliable).

I and koji setup the measurement of the QPD response to the pitch and yaw displacements of the beam spot.

We did this using a 100mW 1064nm laser. Its power was attenuated to ~ 1.9mW, and the spot size at the QPD position was 6000-7000 um .

The QPD was put on a translation stage, using which, the center of teh QPD wrt the beam spot could be moved in pitch and yaw.

Following are the measurements :

For yaw

:

The slope of teh linear region is -8356 /inch

 For pitch

The slope of the linear region in this is 9085/inch

 The old plots looked horrible, and so here is a new plot

The slopes and other stats are

Pitch

Linear model Poly1:
     f(x) = p1*x + p2
Coefficients (with 95% confidence bounds):
       p1 =        8550  (7684, 9417)
       p2 =       -2148  (-2390, -1906)

Goodness of fit:
  SSE: 9944
  R-square: 0.9923
  Adjusted R-square: 0.9907
  RMSE: 44.59

Yaw

Linear model Poly1:
     f(x) = p1*x + p2
Coefficients (with 95% confidence bounds):
       p1 =       -8310  (-8958, -7662)
       p2 =        2084  (1916, 2252)

Goodness of fit:
  SSE: 6923
  R-square: 0.9954
  Adjusted R-square: 0.9945
  RMSE: 37.21

For yesterday - July 12th.

Yesterday, I tried understanding the MEDM and the Dataviewer screens for the WFS.

I then also decided to play around with the sensing matrix put into the WFS control system and see what happens.

I changed the sensing matrix to completely random values, and for some of the very bad values, it even lost lock :P (i wanted that to happen)

Then I put in some values near to what it already had, and saw things again.

I also put in the matrix values that I had obtained from my DC calculations, which after Rana's explanation, I understand was silly.

Later I put back the original values, but the MC lock didnot come back to what it was earlier. Probably my changing the values took it out of the linear region. THE MATRIX NOW HAS ITS OLD VALUES.

I was observing the POwer Spectrum of teh WFS signals after changing the matrix values, but it turned out to  be a flop, because  I had not removed the mean while measuring them.  I will do that again today, if we obtain the lock again (we suddenly lost MC lock badly some 20 minutes ago).

Summary of this week's work Wednesday - Aligned the mode cleaner with Koji, and then measured the beam characteristics at MC2 end. Koji then taught me how to read the WFS signals Thursday - wrote a script to measure the signals and calculated the coefficients relating mirror movement and DC signals of WFS. To know the possibility of the control, found SVD of the coeff matrix, and condition number. Friday - Set up the measurement of QPD linear response using a laser outside the cavity. Took data. Monday - did the calculations and plotting for the above experiment. Then played around with the MEDM screens , and also tried to see what happens to the Power Spectrum of WFS signals by changing the coefficients in teh matrix. (failed) Tuesday - played around with WFS, tried seeing what it does when switched on at different points, and also what it does when I disturb the system while WFS has kept it locked.

I tuned the gain of WFS to 0 last night at about 3am.

I turned it back on now.

Yesterday I installed teh QPD on the table behind MC2, and observed teh signal on it.

The MC_leak is directed to it by a steering mirror.

I used the A2L_MC2 script to minimise  teh pitch and yaw gains, and estimated teh spot position on teh MC2 using that.

This spot position was aligned to the center of teh QPD.

In the night while before taking measurements, I decided to turn off the Wavefront Sensor Servos, but just after that, the MC alignment went very bad, and I could not align it in the next 2 hours.

For some reason, the MC was really mad the whole day yesterday, and was getting misaligned again and again, even when the WFS feedback was on.

The table also had another IR laser in it, which I and Koji switched off.

I will continue measuring once we pump down again.

For now, I am analysing teh QPD circuit Transfer Function.

I turned the WFS gain to 0.02 back, and the MC is locked, the data for the seismic motion might be meaningful nowforth.

Yesterday, I started twiddling with the Mode Cleaner at about 2 pm.

So the seismic data should be all good before that.

I was using it till about 3.30 am, and then left for the night with locking it and swithcing on back the WFS control

Today morning, I have started twiddling with it again, at about 10.30 am.

About my work with the mode cleaner :

I am primarily exciting the mirrors in pitch and yaw, and trying to measure the response of the WFS and the MC2 OPLEV wrt the excitation.

This thus involves switching off the WFS control while measurement.

After two more of those measurements today, I will get to finding new values for the Output Matrix of the WFS for controlling MC1 & 3, and also, try giving in control to MC2 alignment using OPLEV signals.

 TFs after the measurement -

 In the order - MC1 , MC2 , MC3 -pitch and yaw.

These plots let us know about how do the wavefront sensor signals actually respond to the mis-alignments in the mirrors.

For legibility, legend has been includded in only one plot in each pdf., its typically the same for all  3 plots.

the actual xml files for this measurement are in the directory /cvs/cds/caltech/users/nancy/Align_Matrix/highpower/spot_center

It was made sure before each measurement that the MC is best aligned, the WFS are turned off, and the spots on all 3 QPDs are centered.

I calculated the MC1&3 Vs WFS1&2 Output Matrix today from the above measurements with koji's help.

the matrix can be generated from the m file at /cvs/cds/caltech/users/nancy/Align_Matrix/matrix.m

these values were put in, and the direction of control is sort of confirmed. I tried twiddling with the gains in the loop to get a 4*4 stable control, but could not succeed.

the mode cleaner is back locked now, and WFS matrix as well as gains are reverted to the old values.  (1.30 am)

The output Matrices are

Pitch

Yaw

 I realised today morning that there was a flaw in my calculations for the yaw matrix.

Correcting the values, and also making teh tables more readable.

I will test these values once our computers are back to working condition.

 E-log entry for Friday - will attach more plots to this entry on wednesday after i am back   to the 40.

Status	MC_Trans	MC_REF
New Control ON
New Control OFF
Old Control ON

 Well, the data is kind of not enough to be analysed in time domain,

But by far from what I analyse, I think that the new control is not worse than the old one.

I donot also find any better results, except for this one being theoritically stronger.

Since the laser is off, Jenne and I rasied the chiller-chiller (small AC in the Control Room) set point temperature to 73 degree F (from 68F) to save people from shivering.

 I have taken the charger for the dark gray dell laptop from its station, and have labelled the information there too.

Will keep it back tonight.

Friday night myself and Koji measured the Transfer function of the QPD circuit at MC2 side using a chopper . Following was our procedure :

We connected some wires at the input and output of the filter circuit to one of the segment of teh QPD. - seg 1.

A laser light was shined on to the QPD, it was pulsed using a chopper. The frequency of rotation of the chopper was varied.

These wires were then fed to the spectum analyser , and a transfer funstion was observed, It was nearly a low pass filter

The chopper frequency was then made variable by giving the chopper a signal from the spectrum analyser. This signal just swiped a large range of the rpm of the chopper.

Now the input signal looked like a sine wave of varying frequency. the transfer function looked like a perfect LPF, with a small SNR.

Attaching the plot of the TF in the next e-log (this one is on windows and can't access /cvs/cds)

 Yesterday , I put in the Output Matrix, and changed the gain sliders for the 4 WFS loops.

It worked and was keeping the lock for the MC.

I then tested whether the MC1 and 3 were following any change in MC2 alignment. It was indeed workinng,

Next we stepped to putting in the gains for the MC2 oplev servo.

the signs are decided on the basis of convergence, and the magnitude is kept very low, to have a very slow control for MC2.

This complete 6 * 6 model does work, and was able to keep the transmission held.

I also tried poking each mirror in pitcg and yaw, and the cavity comes back to high resonance after some time.

This time is indeed large if the poking is made for MC2, and the transmission comes back to normal after big oscillations.

I tried to measure the Open loop TFs for all these loops yesterday, but somehow could not find a correct excitation.

I will do it today.

Plan ahead :

1.  Center the spot on MC2 and the QPD

2. Optimize the gains by looking at response to noise.

3. Measure Power Spectrum Density of each error signal.

EDIT: The script and template file have been moved to /opt/rtcds/caltech/c1/scripts/general/netgpibdata/

___________________________________________________________________________________________________________

The NEW and IMPROVED script for remotely getting data from Agilent 4395A network analyzer is located at /users/nichin

This script is quite different from the one in Elog 10108 and fetches us both magnitude and phase. There is an added feature of setting the IF Bandwidth.

The network analyzer is located at crocetta.martian (192.168.113.108)

How to run the script:

> python NWAG4395A_data_acq.py [filename.yml]

1. The script accepts sweep parameters and output options via a .yml file that is written following a template that can be found at /users/nichin/NWAG4395A_parameters.yml
2. The data obtained is stored as a .dat file and the corresponding details regarding the acquired data is in a .par parameter file
3. Separate .dat and .par files are created for phase and magnitude of voltage data.
4. You can choose to get a plot of the data obtained by specifying it in the .yml file. The plots are automatically stored as PDF.
5. Plots, data and parameter files are all stored in a new folder that is created with a timestamp in its name.
6. NOTE: Plotting options are only available in computers running numpy versions of 1.6.0 or above.(Currently only Ottavia and Chiara)

Test Run:

I connected a simple 2MHz Low pass filter between the modulation output and signal input of the NA and ran a scan from 100KHz to 20MHz. The script was run from Ottavia.

The expected plot was obtained and is attached here.

Current work:

Setting up the RF switch in rack 1Y1 to select between required PDs and scripts to tell it which channel to choose over the Ethernet.

Yes, at 45degrees PR3 will only have a curvature of about 850m for the vertical mode of the beam, apparently not enough to stabilize the cavity.

 Also make sure you are taking into account the substrate of the ETM.

Is this reflection measured with the cavity locked or unlocked?

So what's the actual designed reflectivity of the ETM for green? No one seems to be able to give me a straight answer about this.

Looking at the reflected beam when the beam is misaligned makes it look like it's << 0.9. Is that expected given the coating spec?

You say the cavity scan goes as high as 300cts but you can only lock to 30cts, are you locked on the sideband?

I'd repeat the measurement for REFL11. The PRC arrow has some big error bar on it, and maybe the true error is even bigger.

Also, please make the placement of the plots the same for modeled and measured so it's easy to compare.

 The 785 analyzer in the 40 had a wonky hard to read screen. I was hoping that a new white CRT would fix all the problems. 

I installed a white CRT, which didn't fix the wonkyness, but I adjusted the CRT position, brightness, focus settings to make the screen somewhat more readable.

BEFORE:

AFTER:

If we want to send the thing in for service to fix the wonkyness, we should probably hold on to the old CRT because they will probably replace the whole screen assembly and we'll lose our white screen.

In order to better understand how the composite signal would behave in the presence of noise, I decided to do a simple analysis of the cavity signals while sweeping through resonance.

My noise model was to just assume that a given signal has some rms uncertainty (error bars) and use linear error propagation to propagate from simple signals to more complicated ones.

I used the python package uncertainties to do the error propagation.

I assumed that the ALS signal, the cavity transmission, and the cavity PDH error signal all have some constant noise that is independent of the cavity detuning. Below is a sweep through resonance (x axis is cavity detuning in units of radians).

The shaded region represents the error on each signal.

Next I calculated the 'first order' calculated error signals. These being a raw PDH, normalized PDH, an inverse square root trans, and the normal ALS again. I tuned the gains so they match appropriately.

Here, one can see how the error in the trans signal propagates to the normalized and trans signals and becomes large are the fractional error in the trans signal becomes large.

Next I did some optimization of linear combinations of these signals. I told the code to maximize the total signal to noise ratio, while ensuring that the overall signal had positive gain. I did this again as a function of the cavity detuning.

Each curve represents the optimized weight of the corresponding signal as a function of detuning.

So this is roughly doing what we expect, it prefers ALS far from the resonance, and PDH close to the resonance, while smoothly moving into square root trans in the middle.

It's a little fake, but it gives us an idea of what the 'best' we can do is.

Finally I used these weights to recombine the signals into a composite, to get an idea of the noise of the overall signal. At the same time, I plot the weighting proposed by Koji's mathematica notebook (using trans and 1-trans, and a hard switch to ALS).

So as one can see, at least for the noise levels I chose, the koji weighting is not much worse than the 'optimal' weighting. While it is much simpler.

The code for all this is in the svn at 40mSVN/nicolas/workspace/2014-03-06_compositeerror

#!/bin/bash   ### this is a comment  PATH=$PATH:/home/user/path    echo "How cool is this?" 

EDIT: it looks like the code highlighting is sort of broken :-(.