40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  40m Log, Page 65 of 344  Not logged in ELOG logo
ID Date Author Type Categoryup Subject
  17017   Tue Jul 19 07:34:46 2022 AnchalUpdateCalibrationError propagation to astrophysical parameters from detector calibration uncertainty

Addressing the comments as numbered:

  1. Yeah, that's correct, that equation normally \Delta \Theta = -\mathbf{H}^{-1} \mathbf{M} \Delta \Lambda but it is different if I define \Gamma bit differently that I did in the code, correct my definition of \Gamma to :
    \Gamma_{ij} = \mu_i \mu_j \left( \frac{\partial g}{\partial \mu_i} | \frac{\partial g}{\partial \mu_j} \right )
    then the relation between fractional errors of detector parameter and astrophysical parameters is:
    \frac{\Delta \Theta}{\Theta} = - \mathbf{H}^{-1} \mathbf{M} \frac{\Delta \Lambda}{\Lambda}
    I prefer this as the relation between fractional errors is a dimensionless way to see it.
  2. Thanks for pointing this out. I didn't see these parameters used anywhere in the examples (in fact there is no t_c in documentation even though it works). Using these did not affect the shape of error propagation slope function vs frequency but reduced the slope for chirped Mass M_c by a couple of order of magnitudes.
    1. I used the get_t_merger(f_gw, M1, M2) function from Hang's work to calculate t_c by assuming f_{gw} must be the lowest frequency that comes within the detection band during inspiral. This function is:
      t_c = \frac{5}{256 \pi^{8/3}} \left(\frac{c^3}{G M_c}\right)^{5/3} f_{gw}^{-8/3}
      For my calculations, I've taken f_{gw} as 20 Hz.
    2. I used the get_f_gw_2(f_gw_1, M1, M2, t) function from Hang's work to calculate the evolution of the frequency of the IMR defined as:
      f_{gw}(t) = \left( f_{gw0}^{-8/3} - \frac{768}{15} \pi^{8/3} \left(\frac{G M_c}{c^3}\right)^{5/3} t \right)^{-3/8}
      where f_{gw0} is the frequency at t=0. I integrated this frequency evolution for t_c time to get the coalescence phase phi_c as:
      \phi_c = \int^{t_c}_0 2 \pi f_{gw}(t) dt
  3. In Fig 1, which representation makes more sense, loglog of linear axis plot? Regarding the affect of uncertainties on Tidal amplitude below 500 Hz, I agree that I was also expecting more contribution from higher frequencies. I did find one bug in my code that I corrected but it did not affect this point. Maybe the SNR of chosen BNS parameters (which is ~28) is too low for tidal information to come reliably anyways and the curve is just an inverse of the strain noise PSD, that is all the information is dumped below statistical noise. Maybe someone else can also take a look at get_fisher2() function that I wrote to do this calculation.
  4. Now, I have made BBH parameters such that the spin of the two black holes would be assumed the same along z. You were right, the gamma matrix was degenerate before. To your second point, I think the curve also shows that above ~200 Hz, there is not much contribution to the uncertainty of any parameter, and it rolls-off very steeply. I've reduced the yspan of the plot to see the details of the curve in the relevant region.

1. In the error propogation equation, it should be \Delta \Theta = -H^{-1} M \Delta \Lambda, instead of the fractional error. 

2. For the astro parameters, in general you would need t_c for the time of coalescence and \phi_c for the phase. See, e.g., https://ui.adsabs.harvard.edu/abs/1994PhRvD..49.2658C/abstract.

3. Fig. 1 looks very nice to me, yet I don't understand Fig. 3... Why would phase or amplitude uncertainties at 30 Hz affect the tidal deformability? The tide should be visible only > 500 Hz.

4. For BBH, we don't measure individual spin well but only their mass-weighted sum, \chi_eff = (m_1*a_1 + m_2*a_2)/(m_1 + m_2). If you treat S1z and S2z as free parameters, your matrix is likely degenerate. Might want to double-check. Also, for a BBH, you don't need to extend the signal much higher than \omega ~ 0.4/M_tot ~ 10^4 Hz * (Ms/M_tot). So if the total mass is ~ 100 Ms, then the highest frequency should be ~ 100 Hz. Above this number there is no signal.


  17029   Sun Jul 24 08:56:01 2022 HangUpdateCalibrationError propagation to astrophysical parameters from detector calibration uncertainty

Sorry I forgot to put tc & phic in the example. 

I modified astroFisherLib.py to include these parameters. Please note that their meaning is that we don't know when the signal happens and at which phase it merges.

It does not mean the time & phase from a reference frequency to the merger. This part is not free to vary because it is fixed by the intrinsic parameters.  

It might be good to have a quick scan through the Cutler & Flanagan 94 paper to better understand their physical meanings.


  17221   Wed Nov 2 16:34:56 2022 PacoSummaryCalibrationSingle arm calibration run

[Anchal, Paco]

We added a notch filter on ETMY (the actuation point of the YARM control loop) to inject our calibration line at 575.170 Hz. The excitation is injected using the DARM Oscillator, with an exc. gain of ~ 500 (this gets us a decent > 10 SNR line in the ALS Y beat). With the arm cavity locked to the PSL (~150 Hz control bandwidth), and the aux laser locked to the cavity (~10 kHz control bandwidth) the goal of this run is to calibrate our actuator strength and more importantly to budget its uncertainty. For this we have looked at the ALS beat stability using Allan statistics; we noticed the ALS beatnote frequency fluctuations start to become dominated by 1/f (or divergent noise due to systematic drifts in the YARM loop) after 10 seconds (see Attachment #1) (we have managed to see 30 seconds stability with the HEPAs off and without locking to IMC).

Our prediction is that our demodulated calibration lines will display the least residual rms noise when averaging down to around this time. This is the only reason one would use allan statistics; to quantify the separation between statistical and systematic effects in a frequency measurement. To be continued...


  • 1351464997 to 1351467139
  • 1351467259 to 1351468221
  • 1351468318 to ...
  17253   Thu Nov 10 17:40:31 2022 PacoSummaryCalibrationCalibration Plan

Plan to calibrate single arm actuation strength

  1. Lock single arm cavity (e.g. YARM)
  2. Lock YAUX laser to arm cavity (actuation point is ETMY)
  3. With the notch on the YARM loop filter (actually on ETMY),
  4. Turn on cal line (e.g. DARM osc) to move ITMY; here the frequency is chosen to be away from 600 Hz (line harmonic) and from violin modes for ITMY (642 Hz). The lower value of 575.17 Hz was chosen to avoid demodulating noise peaks at 455 Hz and 700 Hz.
  5. Get raw YALS beatnote (we chose the demod angle of -35 deg to minimize Q).

The analysis is as follows:

  1. Get demodulated IQ timeseries for the duration of the locks before lowpass filter (C1:CAL-SENSMAT_DARM_BEATYF_I_DEMOD_I_IN1); we are also storing the raw beatnote if we want to do software demodulation.
  2. Look at the allan deviation of I and Q to establish the timescale over which our measurement is dominated by statistical uncertainty -- after this time, the uncertainty is expected to be due to systematic error / drift. In this case as shown by Attachment #1 the time is around 60.6 seconds.
  3. At this frequency and with 500 gain the ITMY coils should be actuating 7.32 pm of amplitude displacement.
  4. The minimum allan deviation does indeed predict the statistical uncertainty limited rms if we look at the power spectra of the demodulated cal line over different time periods (Attachment #2), notice I lowpassed the raw timeseries.
  5. I think the next step is to get the nominal calibration value and repeat the measurment for more than a single cal line.
  6. Roughly from the deviation plot, our fractional beatnote deviation is a proxy for the calibration uncertainty. 1.15e-16 of beatnote stability should translate to a fractional displacement stability of ~4.57e-15 at 60 seconds; giving an ultimate statistical calibration uncertainty of 0.06% at this particular frequency when averaging for this long. It might be interesting to see a calibration frequency dependent allan deviation plot.
  17257   Fri Nov 11 14:15:45 2022 PacoSummaryCalibrationSingle arm cal with 5 lines

I turned all the LSC oscillators on and used the digital demod for BEATYF (fine y als beat) to grab the data. For this I added notches onto SCY ETMY LSC filter banks FM6-10 to account for these lines at 30.92, 211.10, 313.31, 315.17 Hz (basically just reusing the osc models) and adjusted the sensing matrix to actuate on ITMY.

I aligned and locked YARM, and then I aligned and locked the YAUX. The lock seems pretty robust with an avg green transmission GTRY ~ 0.185 counts for TEM00.

Trying to see other lines appear on the BEATYF demod channels, but so far no luck. I scaled down the exc gain from 500 counts (snr ~ 20 at 575 Hz) and verified the notches are working. Since I am unsure of the issue here and WFS tests are happening at 4 today, I decided to take some beat data under different conditions -->

HEPA OFF and PSL Shutter Open

gpstime start = 1352244763 gpstime end = 1352245405


PSL Shutter Closed

gpstime start = 1352245574

gpstime end = 1352246216


PSL Shutter Closed

gpstime start = 1352246240 gpstime end = 1352246882
  17266   Tue Nov 15 11:04:01 2022 PacoSummaryCalibrationSingle arm cal with 5 lines

YALS Hardware inspection

The ALS Cal for ITMY actuation was off by ~ 1000, so I decided I don't trust / understand what this beatnote is seeing. Then, I went in the lab to inspect things;

  1. ALS DFD and DEMOD: The demod box was off no perhaps on purpose, more likely by accident... the switch was off in the rear of the 1U box in the LSC rack... Also, the DB9 cable labeled ALS was disconnected. Fixed both these bugs and verified it worked all the way into cds.
  2. YARM Green injection: Did some re-alignment, and noted the mode was hopping a bit too much (once every couple of seconds) so I rotated the half waveplate before the green PZT steering mirrors by  ~ 8 deg and used the latter to get a GTRY transmission of > 0.320 (counts), about 75% more than last time. Finally I made sure the mode is robustly locked for several minutes.

Beatnote calibration

The factor above may be explained by the bogus signals coming into the beat fine phase channels on the ALS model. After locking the YARM with POY11, and locking the YAUX to the YARM cavity, I turned on the LSC oscillators -- all five of them see Attachment #1 for the screenshot -- and looked for the lines in the C1:ALS-BEATY_FINE_PHASE_OUT channel. Here, again the sensing output matrix was set up to actuate on ITMY, while the ETMY (control point of YARM loop) had all the output notches on. Once all lines were visible in the YAUX beatnote, I had to reduce the LSC filter gain from 0.012 to 0.011 to prevent loop oscillations... Then I recorded the gpstimes below with different conditions.

  • HEPA ON (JC Inside lab)
    • gpstime = 1352575798 to 1352576046
  • HEPA OFF (JC Inside lab)
    • gpstime = 1352576125 to 1352576216
  • HEPA OFF (JC in the control room)
    • gpstime = 1352576641 to 1352577591


Basically, only the DARM line was recorded (DQ channs) so I modified the c1cal to store the SIG_OUT and DEMOD_I_IN1 channels for both BEATX and BEATY cal signals. This means I need to repeat this measurement. In the meantime I am also going to try and rerun calibrate the BEAT HZ transfer function.

  17291   Mon Nov 21 11:52:50 2022 RadhikaSummaryCalibrationSingle arm cal with 5 lines

[Paco, Radhika]

We set out to realign the YARM AUX laser input into the arm cavity.

- We noticed that the GTRY beam was way off the center of the screen, so we went to the vertex table to align the camera.

- The beam spot at GTRY PD was large/divergent, so we shifted the PD closer to the penultimate mirror. We also doubled the PD gain. Transmission went from ~0.3 to ~0.7 (with gain doubled).

- We returned to the YARM end table to finalize alignment with the green PZT steering mirrors. GTRY was maximized to ~0.77.

  17328   Wed Nov 30 20:01:08 2022 ranaSummaryCalibrationSingle arm cal with 5 lines

I don't think you need to record the excitations. They are just sine waves. The amplitude you can read off from the OSC screen. You just have to have the BEAT channel recorded and you can demod it to get the calibration. If the BEAT channel is calibrated in Hz, and you know the 40m arm length, then you're all done.



Basically, only the DARM line was recorded (DQ channs) so I modified the c1cal to store the SIG_OUT and DEMOD_I_IN1 channels for both BEATX and BEATY cal signals. This means I need to repeat this measurement. In the meantime I am also going to try and rerun calibrate the BEAT HZ transfer function.


  17329   Thu Dec 1 20:43:25 2022 AnchalSummaryCalibrationSingle arm cal with 5 lines

[Anchal, Paco]

We are doing this attempt again in following configuration:

  • PSL shutter is closed. (So IR laser is free running)
  • Beanote frequency between Y arm and Main laser is about 45 MHz.
  • Green laser on Y end is locked. Transmission is above 1.1 (C1:ALS-TRY_OUT)
  • All calibration oscillators are turned on and set to actuate ITMY. See screenshot attached.
  • The calibration model was changed to demodulate the C1:ALS-BEATY_FINE_PHASE_OUT channel insteald. We'll have DQ channels before mixing with oscillator, after mixing, and also after applying a 4th order 30 Hz butterworth filter.

Start time:

PST: 2022-12-01 20:44:23.982114 PST
UTC: 2022-12-02 04:44:23.982114 UTC
GPS: 1353991481.982114

Stop time:

PST: 2022-12-02 14:32:29.547603 PST
UTC: 2022-12-02 22:32:29.547603 UTC
GPS: 1354055567.547603

  17346   Thu Dec 8 16:21:40 2022 PacoSummaryCalibrationITMY actuation strength cal with 5 lines

[Anchal, Paco]

After debugging the hardware, on gpstime 1354422834 we turned on 5 cal lines on ITMY to test the ALS calibration for the single arm along with our error estimates.

Note: the YARM IR lock lasted > 8 hours, but the GRY transmission dropped twice during the evening and hopped back up, so the phase tracker jumped a couple of times.


YAUX laser was locked to the YARM through the analog PDH servo (UGF ~ 2 kHz), YARM was locked to the PSL with POY11 (UGF ~ 200 Hz), and the ALS phase tracker was set to output the beat frequency noise in Hz. HEPA was left on during this measurement. The oscillators were similar to previous instances: gains of 70@211.1Hz, 100@313.31Hz, 100@315.17Hz, 300@575.17Hz and 15@30.92Hz with appropriate notches on ETMY to avoid POY11 loop supression.


For YARM, the high bandwidth YAUX laser loop with transfer function G ensures that the relative laser frequency fluctuations correlate with the relative length fluctuations as:

\frac{\delta \nu}{\nu} = \frac{G}{1 - G} {\frac{\delta L}{L}}

Then, getting the magnitude of the YARM displacement at calibration frequencies is possible by knowing the arm cavity length, open loop gain, and absolute frequency (wavelength). The relative calibration error on the magnitude of the displacement is

\frac{\Delta \lvert \delta L\lvert}{\lvert \delta L \lvert} = \left[ \left(\frac{\Delta {L}}{L}\right)^2+ \left(\frac{\Delta {\lambda}}{\lambda}\right)^2 + \left(\frac{\Delta {{\delta \nu}}}{{\delta \nu}}\right)^2 + \left(\frac{\Delta \lvert G \lvert}{\lvert G \lvert(\lvert G \lvert - 1)}\right)^2 \right]^{1/2}

including the relative uncertainties in the YARM length, wavelength, and open loop gain. Interestingly, the loop gain term weighs proportionally less as G increases, so even if G = 100 (10), its relative error contribution would be < 1%. To estimate our total error, we assume the wavelength and YARM length are 1064.1(5) and 37.79(1), and add the frequency dependent values for G with 10% error. Finally, we use the rms ASD to estimate the relative error from the beatnote fluctuation measurement.

The measurement was done similar to other instances, taking the 'C1:ALS-BEATY_FINE_PHASE_OUT_HZ_DQ' timeseries (sampled at 16 kHz) and demodulating at the calibration frequencies above to get the mean YAUX laser frequency fluctuation and its uncertainty from the demodulated rms ASD.


Attachment #1 shows the raw timeseries, Attachment #2 shows the spectra around the cal lines, Attachment #3 shows the demodulated timeseries, Attachment #4 shows the final result for the 5 lines, including the tallied errors as detailed above.

ITMY actuation = 4.92(11) nm / count / f^2


We compared our results from Attachment #4 against a MICH referenced ITMY actuation calibration found here; which Yuta guess-timated a 10% uncertainty (gray shaded band in Attachment #4). An important correction came for the 575 Hz line, not just because the YAUX OLG is small but because a violin filter on ITMY LSC output has a 1.4475 gain bump. In fact we collected any additional digital gains from the ITMY output filters:

Output filter digital gains at cal lines (from foton)
30.92 Hz 211.1 Hz 313.31 Hz 315.31 Hz 575.17 Hz
1.00007 1.0034 1.0101 1.01017 1.4475


  • Consider moving the 575 Hz line to avoid additional digital gain, but try to remain at high frequency.
    • Maybe here we can use the resonant gain MOKU filters that Radhika is designing.
  • Setup a live loop gain calibration to reduce the uncertainty for the high frequency cal line.
    • We can also just grab GTRY (transmission) as a proxy for optical gain and use for budgeting.
  • Work on setting up constraints for error mitigation based on allan deviation of the beatnote and PDH nonlinearity.
  • Move this to FPMI or some other lock configuration
  10436   Thu Aug 28 11:02:53 2014 SteveUpdateCalibration-RepairSR785 repair

SN 46,795 of 2003 is back.

  11641   Thu Sep 24 17:06:14 2015 ericqUpdateCalibration-RepairC1CAL Lockins

Just a quick note for now: I've repopulated C1CAL with a limited set of lockin oscillators/demodulators, informed by the aLIGO common LSC model. Screens are updated too. 

Rather than trying to do the whole magnitude phase decompostion, it just does the demodulation of the RFPD signals online; everything beyond that is up to the user to do offline. 

Briefly testing with PRMI, it seems to work as expected. There is some beating evident from the fact that the MICH and PRCL oscillation frequencies are only 2Hz apart; the demod low pass is currently at an arbitrary 1Hz, so it doesn't filter the beat much. 

Screens, models, etc. all svn'd.

  12040   Mon Mar 21 14:29:32 2016 SteveUpdateCalibration-Repair1W Innolight laser repair diagnoses



After adjusting the alignment of the two beams onto the PD, I managed to recover a stronger beatnote of ~ -10dBm. I managed to take some measurements with the PLL locked, and will put up a more detailed post later in the evening. I turned the IMC autolocker off, turned the 11MHz Marconi output off, and closed the PSL shutter for the duration of my work, but have reverted these to their nominal state now. The are a few extra cables running from the PSL table to the area near the IOO rack where I was doing the measurements from, I've left these as is for now in case I need to take some more data later in the evening...I

Innolight 1W 1064nm, sn 1634 was purchased in 9-18-2006 at CIT. It came to the 40m around 2010

It's diodes should be replaced, based on it's age and performance.

RIN and noise eater bad. I will get a quote on this job.

The Innolight Manual frequency noise plot is the same as Lightwave' elog 11956

Diagnoses from Glasglow:

“So far we have analyzed the laser. The pump diode is degraded. Next we would replace it with a new diode. We would realign the diode output beam into the laser crystal. We check all the relevant laser parameters over the whole tuning range. Parameters include single direction operation of the ring resonator, single frequency operation, beam profile and others. If one of them is out of spec, then we would take actions accordingly. We would also monitor the output power stability over one night. Then we repackage and ship the laser.”

  12045   Thu Mar 24 07:56:09 2016 SteveUpdateCalibration-RepairNO Noise Eater for 1W Innolight

1W Innolight is NOT getting Noise Eater as it was decided yesterday at the 40m meeting. Corrected 3-25-2016

Repair quote with adding noise eater is in 40m wiki




After adjusting the alignment of the two beams onto the PD, I managed to recover a stronger beatnote of ~ -10dBm. I managed to take some measurements with the PLL locked, and will put up a more detailed post later in the evening. I turned the IMC autolocker off, turned the 11MHz Marconi output off, and closed the PSL shutter for the duration of my work, but have reverted these to their nominal state now. The are a few extra cables running from the PSL table to the area near the IOO rack where I was doing the measurements from, I've left these as is for now in case I need to take some more data later in the evening...I

Innolight 1W 1064nm, sn 1634 was purchased in 9-18-2006 at CIT. It came to the 40m around 2010

It's diodes should be replaced, based on it's age and performance.

RIN and noise eater bad. I will get a quote on this job.

The Innolight Manual frequency noise plot is the same as Lightwave' elog 11956

Diagnoses from Glasglow:

“So far we have analyzed the laser. The pump diode is degraded. Next we would replace it with a new diode. We would realign the diode output beam into the laser crystal. We check all the relevant laser parameters over the whole tuning range. Parameters include single direction operation of the ring resonator, single frequency operation, beam profile and others. If one of them is out of spec, then we would take actions accordingly. We would also monitor the output power stability over one night. Then we repackage and ship the laser.”


  12070   Mon Apr 11 17:03:41 2016 SteveUpdateCalibration-Repair1W Innolight repair completed

The laser is back. Test report is in the 40m wiki as New Pump Diode Mephisto 1000

It will go on the PSL table.

  13456   Tue Nov 28 17:27:57 2017 awadeBureaucracyCalibration-RepairSR560 return, still not charging

I brought a bunch of SR560s over for repair from Bridge labs. This unit, picture attached (SN 49698), appears to still not be retaining charge. I’ve brought it back. 

  14759   Mon Jul 15 03:30:47 2019 KruthiUpdateCalibration-RepairWhite paper as a Lambertian scatterer

I made some rough measurements, using the setup I had used for CCD calibration, to get an idea of how good of a Lambertian scatterer the white paper is. Following are the values I got:

Angle (degrees) Photodiode reading (V)  Ps (W) BRDF (per str) % error
12 0.864 2.54E-06 0.334 20.5
24 0.926 2.72E-06 0.439 19.0
30 1.581 4.65E-06 0.528 19.0
41 0.94 2.76E-06 0.473 19.8
49 0.545 1.60E-06 0.423 22.5
63 0.371 1.09E-06 0.475 28

Note: All the measurements are just rough ones and are prone to larger errors than estimated.

I also measured the transmittance of the white paper sample being used (it consists of 2 white papers wrapped together). It was around 0.002

  14804   Wed Jul 24 04:20:35 2019 KruthiUpdateCalibration-RepairMC2 pitch and yaw calibration

Summary:  I calibrated MC2 pitch and yaw offsets to spot position in mm. Here's what I did:

  1. Changed the MC2 pitch and yaw offset values using  ezca.Ezca().write('IOO-MC2_TRANS_PIT_OFFSET', <pitch offset value> ) and ezca.Ezca().write('IOO-MC2_TRANS_YAW_OFFSET', <yaw offset value> )
  2. Waited for ~ 700-800 sec for system to adjust to the assigned values
  3. Took snapshots with the 2 GigEs I had installed - zoomed in and zoomed out. (I'll be using these to make a scatter loss map, verify the calibration results, etc)
  4. Ran the mcassDecenter script, which can be found in /scripts/ASS/MC. This enters the spot position in mm in the specified text file.

Results:  In the pitch/yaw vs pitch_offset/yaw_offset graph attached,

  • intercept_pitch = 6.63 (in mm) ,  slope_pitch = -0.6055 (mm/counts) 
  • intercept_yaw = -4.12 (in mm) ,  slope_yaw = 4.958 (mm/counts) 
  15510   Sat Aug 8 07:36:52 2020 Sanika KhadkikarConfigurationCalibration-RepairBS Seismometer - Multi-channel calibration

Summary : 

I have been working on analyzing the seismic data obtained from the 3 seismometers present in the lab. I noticed while looking at the combined time series and the gain plots of the 3 seismometers that there is some error in the calibration of the BS seismometer. The EX and the EY seismometers seem to be well-calibrated as opposed to the BS seismometer.

The calibration factors have been determined to be :

BS-X Channel: \small {\color{Blue} 2.030 \pm 0.079 }

BS-Y Channel: \small {\color{Blue} 2.840 \pm 0.177 }

BS-Z Channel: \small {\color{Blue} 1.397 \pm 0.182 }

Details :

The seismometers each have 3 channels i.e X, Y, and Z for measuring the displacements in all the 3 directions. The X channels of the three seismometers should more or less be coherent in the absence of any seismic excitation with the gain amongst all the similar channels being 1. So is the case with the Y and Z channels. After analyzing multiple datasets, it was observed that the values of all the three channels of the BS seismometer differed very significantly from their corresponding channels in the EX and the EY seismometers and they were not calibrated in the region that they were found to be coherent as well. 

Method :

Note: All the frequency domain plots that have been calculated are for a sampling rate of 32 Hz. The plots were found to be extremely coherent in a certain frequency range i.e ~0.1 Hz to 2 Hz so this frequency range is used to understand the relative calibration errors. The spread around the function is because of the error caused by coherence values differing from unity and the averages performed for the Welch function. 9 averages have been performed for the following analysis keeping in mind the needed frequency resolution(~0.01Hz) and the accuracy of the power calculated at every frequency. 

  1. I first analyzed the regions in which the similar channels were found to be coherent to have a proper gain analysis. The EY seismometer was found to be the most stable one so it has been used as a reference. I saw the coherence between similar channels of the 2 seismometers and the bode plots together. A transfer function estimator was used to analyze the relative calibration in between all 3 pairs of seismometers. In the given frequency range EX and EY have a gain of 1 so their relative calibration is proper. The relative calibration in between the BS and the EY seismometers is not proper as the resultant gain is not 1. The attached plots show the discrepancies clearly : 
  • BS-X & EY-X Transfer Function : Attachment #1
  • BS-Y & EY-Y Transfer Function : Attachment #2

          The gain in the given frequency range is ~3. The phase plotting also shows a 180-degree phase as opposed to 0 so a negative sign would also be required in the calibration factor. Thus the calibration factor for the Y channel of the BS seismometer should be around ~3. 

  • BS-Z & EY-Z Transfer Function : Attachment #3

The mean value of the gain in the given frequency range is the desired calibration factor and the error would be the mean of the error for the gain dataset chosen which is caused due to factors mentioned above.

Note: The standard error envelope plotted in the attached graphs is calculated as follows :

         1. Divide the data into n segments according to the resolution wanted for the Welch averaging to be performed later. 

         2. Calculate PSD for every segment (no averaging).

         3. Calculate the standard error for every value in the data segment by looking at distribution formed by the n number values we obtain by taking that respective value from every segment.

Discussions :

The BS seismometer is a different model than the EX and the EY seismometers which might be a major cause as to why we need special calibration for the BS seismometer while EX and EY are fine. The sign flip in the BS-Y seismometer may cause a lot of errors in future data acquisitions. The time series plots in Attachment #4 shows an evident DC offset present in the data. All of the information mentioned above indicates that there is some electrical or mechanical defect present in the seismometer and may require a reset. Kindly let me know if and when the seismometer is reset so that I can calibrate it again. 

  227   Tue Jan 8 15:20:17 2008 PkpUpdateCamerasGigE update
[Tobin , Pinkesh]

Finally we got the camera doing something (other than giving out its attributes). The only thing that seems to work so far is a program called AAviewer, which converts the image into an ASCII format and displays it on the screen. If you want to play around with it, log into mafalda ( via rana.ligo.caltech.edu. Access /cvs/cds/caltech/target/Prosilica/bin-pc/x86/ and there should be a few programs in there, one of which is AAviewer, which requires you to get an IP address (which is for the camera right now. (You can also get the IP information via the ListCameras program). The camera is physically in the 40m near the network rack.

Other programs dont seem to be working and its probably due to the network/packetsize issues. Since linux2 can change its packetsize to a higher number, I will get it to compile on linux2 for now and then give it a shot.
  234   Thu Jan 10 13:45:52 2008 PkpUpdateCamerasGLIBC Error
So, I have tried to compile the camera files which are in /cvs/cds/caltech/target/Prosilica/examples for the past 2 days now and have been unable to get rid of the following error. (specifically ListCameras.cpp, as it doesnt have any other libraries required, which unnecessarily complicates things)

../../bin-pc/x86/libPvAPI.so: undefined reference to `__stack_chk_fail@GLIBC_2.4'
collect2: ld returned 1 exit status
make: *** [sample] Error 1

I used to get this error on mafalda too, but I had fixed it by installing the latest version of the glibc libraries. Inspite of doing so on linux2, the error still persists. I suspect it had something to do with it being a FC3 machine. My own laptop, which also runs Ubuntu works fine too. The problem with these Ubuntu machines is that they dont let me set the packet sizes to 9 kb which is required by the camera. Linux2 does.

If anyone has any idea how to resolve this issue, please let me know.

  236   Fri Jan 11 17:01:51 2008 pkpUpdateCamerasGigE again
So, here I detail all the efforts on the GigE so far

(1) The GiGE camera requires a minimum of 9 kb packet size, which is not available on mafalda or on my laptop ( both of which run Ubuntu and the Camera programs compile there). The programs which require smaller sizes work perfectly fine on these machines. I tried to statically compile the files on these machines so that I could then port them to the other machines. But that fails because the static libraries given by the company dont work.

(2) On Linux2, which lets me set a packet size as high as 9 Kb, it doesnt compile because of a GLIBC error. I tried updating the glibc and it tells me that the version already existing is the latest ( which it clearly is not). So I tried to uninstall GLIBC and reinstall it, but it wont let me uninstall (it == rpm) glibc, since there are a lot of dependencies. A dead end in essence.

Steps being taken

(1) Locally installing the whole library suites on linux2. Essentially install another version of gcc and g++ and see if that helps.
(2) IF this doesnt work, then the only course of action I can take is to cannibalize linux2's GigE card and put it on mafalda. ( I need permission for this Smile ).

Once again any suggestions welcome.
  245   Thu Jan 17 15:11:13 2008 josephbUpdateCamerasWorking on Malfalda
1) I can statically compile the ListCamera code (which basically just goes out and finds what cameras are connected to the network) on Malfalda and use that compiled code to run on Linux2 without a problem. Simply needed to add explicit links to libpthread.a and librt.a.
(i.e. -Bstatic -L /usr/lib/ -lpthread -Bstatic -L /usr/lib -lrt)

With appropriate static libraries, it should be possible to port this code to other linux machines even if we can't get it to compile on the target machine itself.

2)I've modified the Snap.cpp file so that it uses a packet size of 1000 or less. This simply involves setting the "PacketSize" attribute with the built in functions they provide in their library. After un-commenting some lines in that code, I was able to save tiff type images from the camera of up to 400x240 pixels on Malfalda. The claimed maximum resolution for the camera is 752x480, but it doesn't seem to work with the current setup. The max number of pixels seems to about 100 times the packet size. I.e. packet size of 1000 will allow up to 400x240 (96000) but not 500x240 (120,000). Not sure if this is an issue just with snap code or the general libraries used.

3)Will be working towards getting video running over the next day or so.
  266   Fri Jan 25 11:38:16 2008 josephbConfigurationCamerasWorking GiGE video on Linux - sort of
1)I have been able to compile the SampleViewer program which can stream the video from the Prosilica 750C camera. This was accomplished on my 64-bit laptop running Ubuntu, after about 3 hours of explicitly converting strings to wxStrings and back again within the C++ code. (There was probably an easier way to simply overload the functions that were being called, but I wasn't sure how to go about doing so). By connecting it to the CDS network, I was able to immediately detect the camera and display the images.

Unfortunately, I have not yet been able to get it to compile on Mafalda with the x86 architecture. This may be do the fact that it has wxWidgets version 2.8.7 while my laptop has 2.8.4. Certainly the failure at compile time looks different from the errors earlier, and seem to be within the wxWidget code rather than the SampleViewer code. I may simply need to uninstall 2.8.7 and install 2.8.4 of wxWidgets.

The modified code that will compile on my machine has been copied to /cvs/cds/caltech/target/Prosilica/examples/SampleViewer2b.

2)The Snap program (under /cvs/cds/caltech/target/Prosilica/examples/Snap) also will now take full resolution images even on Mafalda. This was achieved by reducing the packet size to 1000 and also increasing the wait until timeout time up to 400 ms, which originally was at 100. Apparently, it takes on the order of 1 ms per packet as far as I can tell. So full resolution at 752x480 required something of order 360 packets.

To Do:
1) Get sample viewer to compile on Mafalda, and then statically compile it so it can be run from any Linux based machine.
2) Get a user friendly version of Snap up and running, statically compiled, with options for a continuous loop every X seconds and also to set desired parameters (such as height, width, file name to save to, save format, etc).
3) Figure out data analysis with the images in Matlab and an after the fact image viewer.

Attached is an example .tiff image from the Snap program.
  267   Fri Jan 25 13:36:13 2008 josephbConfigurationCamerasWorking GiGE video on Mafalda
Finally got the GiGE camera sample viewer video running on Mafalda by updating to the latest API (version 1.16 from Dec 16, 2007) from Prosilica and then using the modified Sample Viewer code I had written. The API version previously in cvs was 1.14.

It can currently be run by ssh -X into Mafalda and going to /cvs/cds/caltech/target/Prosilica/bin-pc/x86 and running the SampleViewer executable found there.
  289   Thu Jan 31 16:53:41 2008 josephbConfigurationCamerasImproving camera user interface
There's a new and improved version of Snap program at the moment people are free to play with.

Located in /cvs/cds/caltech/target/Prosilica/40mCode/

The program Snap now has a -h or --help option which describes some basic command line arguments. The height (in pixels), width (in pixels), exposure time (in micro seconds), file name to be saved to (in .tiff format), and packet size can all be set. The format type (i.e. pixel format such as Mono8 or Mono16) doesn't work at the moment.

At the moment, it only runs on mafalda.

Currently in the process of adding a loop option which will take images every X seconds, saving them to a given file name and then appending the time of capture to the file name.

After that need to add the ability to identify and choose the camera you want (as opposed to the first one it finds).

Lastly, I've been finding on occassion that the frame fails to save. However if you try again a few seconds later with the exact same parameters, it generally does save the second time. Not sure whats causing this, whether on the camera or network side of things.

I've attached two images, the first at default exposure time (15,000 microseconds) and the second at 1/5th that time (3,000 microseconds).
The command line used was "./Snap -E 3000 -F 'Camera_exp_3000.tiff' "
  292   Fri Feb 1 15:04:54 2008 josephbConfigurationCamerasSnap with looping functionality available
New GiGE camera code is available in /cvs/cds/caltech/target/Prosilica/40mCode/. Currently only runs on Mafalda.

Snap has expanded functionality to continuously loop infinitely or for a maximum number of images set by the user. File names generated with the loop option have the current Unix time and .tiff appended to them. So -f './test' will produce tiff files with format "test1234567.tiff". The -l option sets the number of seconds between images.

"./Snap -l 5 -i -f './test' " will cause the program to infinitely loop, saving images every 5 seconds. Using "-m 10" instead of "-i" will take a series of 10 images every 5 seconds (so taking a total of 50 seconds to run).

It also now defaults to 16-bit (in reality only 10 bit) output instead of 8 bit output. You can select between the two with -F 'Mono8' or -F 'Mono16'.

Use --help for a full list of options.

Note that if you ctrl-c out of the loop, you may need to run ./ResetCamera (or whatever the IP is - use ./ListCameras to determine IP if necessary) in order to reset the camera because it doesn't close out elegantly at the moment.
  297   Tue Feb 5 15:32:29 2008 josephbConfigurationCamerasPMC and the GigE Camera
The PMC transmission video camera has been removed and replaced with the GigE GC750 camera for the moment.

A ND4.0 filter has been added in the path to that camera to reduce saturation for the moment.

The old camera has been placed on the elevated section inside the enclosure, and the cable for it is still on the table proper.

The Gige camera is currently running the Snap code on Linux3 with the following command line:

./Snap -E 2000 -l 60 -m 1440 -f './pmc_trans/pmc_trans'

So its going to be taking tiff images every minute for the next 24 hours into the cvs/cds/caltech/target/Prosilica/40mCode/pmc_trans/ directory.

Attached is an example image with exposure set to 2000, loaded into matlab and plotted with the surf command. 2500 microseconds looked like it was still saturating, but this seems to be a good level (with a max of 58560 out of 65535).
  300   Wed Feb 6 16:50:47 2008 josephbConfigurationCamerasRegions of Interest and max frame rate
The Snap code has once again been modified such that setting the -l option to 0 will take images as fast as possible. Also, the -H and -W options set the height and width, while in principle the -Y and -X options set the position in pixels of the top edge and left edge of the image. It also seems possible to set these values such that the saved image wraps around. I'll be adding some command checking so that the user can't do this in the near future.

Doing some timed runs, using a -H 350 and -W 350 (as opposed to the full 752x480), 100 images can be saved in roughly 8 seconds, and 1000 images took about 73 seconds. This corresponds to a frame rate of about 12-13 frames per second (or a 12-13 Hz display). The size of this area was sufficient to cover the current PMC transmission beam.

The command line I used was

time ./Snap -l 0 -m 1000 -f 'test' -W 350 -H 350 -Y 50 -X 350 -E 2000

Interestingly enough, there would be bursts of failed frame saves if I executed commands in another terminal (such as using ls on the directory where the files were being stored).

As always, this code is available in /cvs/cds/caltech/target/Prosilica/40mCode/.
  301   Wed Feb 6 19:39:11 2008 ranaConfigurationCamerasRegions of Interest and max frame rate
We really need to look into making the 40m CDS network have an all GigE backbone so that we can have cooler cameras as well as collect multiple datastreams...
  378   Fri Mar 14 12:06:29 2008 josephbConfigurationCamerasGC750 looking at ETMX while locked
The GC750 (CMOS) is currently looking at the front of ETMX. Unfortunately, its being routed through a 10Mbit connection (which I will be purchasing a replacement for today), so getting it to send images to Mafalda/Linux 2 or 3 isn't working well, but by using a local gigabit switch and a laptop I can get sufficient speed for full images with the sample viewer.

The attached image is from a full 752x480 reslution with 10,000 microsecond exposure with the X-arm locked. Although it looks like I still need to work on the focusing. Will be switching the GC750 with the GC 650 (CCD) later today and comparing the resulting images.
  379   Fri Mar 14 14:59:51 2008 josephbConfigurationCamerasComparison between GC650 (CCD) and GC750 (CMOS) looking at ETMX
Attached are images taken of ETMX while locked.

The first two are 300,000 microsecond exposure time, with approximately the same focusing/zoom. (The 750 is slightly more zoomed in than the 650 in these images). The second are 30,000 microsecond exposures. The la

The CMOS appears to be more sensitive to the 1064 nm reflected light (resulting in bright images for the same exposure time). This may make a difference in applications where images are desired to be taken quickly and repeatedly.

Both seem to be resolving individual specks on the optic reasonably well.

Next test is to place both camera on a Gaussian beam (in a couple different modes say 00, 11, and so forth), probably using the PMC.
  434   Tue Apr 22 08:34:22 2008 josephbConfigurationCamerasCurrent Network Diagram
The attached network diagram has also been added to the 40m Wiki at http://lhocds.ligo-wa.caltech.edu:8000/40m/Image_Processing_with_GigE_Cameras
  471   Thu May 8 16:40:36 2008 josephbConfigurationCamerasGige Camera currently on PSL table
Andrey and myself were working on the PSL table today, using a pickoff of a pickoff of the main beam (adding a microscope slide to pickoff ~4% of the original pickoff) to the GC750 GigeCam.

At the time we left, we scanned the area with a beam scan and didn't see any new stray beams, and nothing in any useful beam paths should have changed. We also strung a Cat 6 cable from the control room switch out to the PSL table in the cable trays, and then above the PSL table.

Currently, its not as well aligned as it could be, and also requires a very low exposure setting, of -E 50 or so to avoid saturation.
  481   Thu May 15 16:24:18 2008 josephbSummaryCameras 
The GC750 camera is currently looking at a very small pickoff of the PSL output (transmission of a Y1-1037-45-S mirror). The plan is to take images tomorrow with it and the GC650 from the same spot and do comparisons.

For those interested, the camera can be run with two codes, from mafalda. Use ssh -X mafalda to login, to allow the live stream to work with the SampleViewer code. The codes can be found in:




Type Snap --help for a list of options for that program. Click the circle looking thing in SampleViewer to start the live stream. Note only 1 of the two programs can be running at a time, and the only way to change settings (such as exposure length) is with Snap at the moment.
  482   Fri May 16 14:38:50 2008 josephbSummaryCamerasTwo cameras setup
I've changed the pickoff setup from yesterday for the GigE cameras to include a 33% beam splitter (first one I could find). The reflection is going to the GC650 (CCD camera) while the transimission is going to the GC750 CMOS camera. This means the CMOS camera has roughly twice the light incident as the GC650 and should be kept in mind in all comparisons. The distances from the beam splitter are approximately the same both cameras, but some more accurate positioning might be useful.

Its very easy to get the GC650 camera into a bad state where you need to go out and cycle the power (simply unplug and re-plug in the power supply either at the camera or outlet). If the ListCamera program doesn't see it, this is probably necessary.

Andrey added at 6.30PM: Actually the 650 camera keeps crashing constantly. Every time I attempt to capture an image, the camera fails.
  506   Fri May 30 12:03:08 2008 josephb, AndreyConfigurationCamerasHead to head comparison of cameras
Andrey and myself - Joseph B. - have examined the output of the GC650 (CCD) and GC750 (CMOS) prosilica cameras. We did several live motion tests (i.e. rotate the turning mirror, move and rotate the camera, etc) and also used a microscope slide to try to eliminate back reflections and interference.

Both the GC650 and GC750 produce dark lines in the images, some of which look parallel, while others are in much stranger shapes, such as circles and arcs.

Moving the GC750 camera physically, we have the spot moving around, with the dark lines appearing to be fixed to the camera itself, and remain in the same location on the detector. I.e. coming back to the same spot keeps showing a circle. In reasonably well behaved sections, these lines are about 10% dips in power, and could in principle be subtracted out. Its possible that the camera was damaged with too much light incident in the past, although going back to the pmc_trans images that were taken, similar lines are still visible.

Moving the GC650 camera physically seems to change the position of the lines (if one also rotates the turning mirror to get to the same spot on the CCD). It seems as if a slight change in angle has a large effect on these dark bands, which can either be thin, or very large, bordering on the size of the spot size. My guess is (as the vendor suggested) the light is interacting with the electronics behind the surface layer rather than a surface defect producing these lines. Using a microscope slide in between the turning mirror and the GC650, we were able to produce new fringes, but didn't affect the underlying ones.

Placing a microscope slide in between the last turning mirror and the GC750 does not affect the dark lines (although it does seem to add some), nor does turning the final turning mirror, so it seems unlikely to be caused by back reflection in this case.

So it seems the CMOS may be more consistent, although we need to determine if the current line problems are due to exposure to too much light at some point in the past (i.e. I broke it) or they come that way from the factory.

Attached are the results of image-processing of the images from the two our cameras using Andrey's new Matlab script.
  511   Mon Jun 2 12:20:35 2008 josephbBureaucracyCamerasBeam scan has moved
The beamscan has been moved from the Rana lab back over to the 40m, to be used to calibrate the Prosilica cameras.
  512   Tue Jun 3 02:15:29 2008 AndreySummaryCamerasFitting results

There have been a lot of work going on related to the processing of images captured by the cameras GC-650 and GC-750 recently.

In the end of the week of May 30 Joseph and me (Andrey) installed the two cameras capturing the images of the pick-off of the main beam on the PSL optical table. The cameras are located after the picked-off beam going towards the "PSL position QPD", after the 33-66 beamsplitter (33% of reflection and 66% of transmission).

Initially (on May 30) the GC-650 camera was taking the images of reflected beam, while the camera GC-750 was taking images of transmitted beam. On Monday June 2 we switched the positions of the cameras, so GC-650 appeared to be on the path of the transmitted beam and GC-750 on the path of the reflected beam.

I (Andrey Rodionov) was able in the weekend to succeed in writing a Matlab program that performs the two-dimensional Gaussian fitting of the captured images, and I used that program to fit the images from the cameras.

The program fits the camera data by a two-dimensional Gaussian surface:

Z = A * exp[ - 2 * (X - X_Shift)^2 / (Waist_X)^2 ] * exp[ - 2 * (Y - Y_Shift)^2 / (Waist_Y)^2 ] + CONST_Shift,

where A, X_Shift, Waist_X, Y_Shift, Waist_Y, CONST_Shift are 6 parameters of the fit.

Attached are the pdf-files showing the results: images taken with our cameras, the 2-dimensional Gaussian fit for these images and the surfaces of residuals. Residuals are differences between the exact beam profile and the result of fitting. In normalized version of residual graph I normalize it by the first coefficient of fitting A, the factor in front of the exponents.
  515   Tue Jun 3 12:33:36 2008 AndreyUpdateCamerasAndrey, Josephb

Continuing our work with cameras,

1) we removed both cameras from their places on Monday afternoon, and were taking the beam-scans with a special equipment (see elog-entry 511) from Bridge bld.,

2) and on Tuesday morning we putted back the GC-750 camera into the transmitted beam path, camera GC-650 into the reflected beam path. We plan to compare the images from the "reflection camera" for several different angles of tilt of the camera.
  517   Wed Jun 4 13:46:42 2008 josephbConfigurationCamerasChanging incident angle images
Attached are images from the GC650 and GC750 when the incident angle was varied from 0 tilt (normal incidence) to 5,10, and 20 degrees. Each time the beam was realigned via the last turning mirror to be on roughly the same spot. This light was a pickoff of the PSL table light just before it leaves the table.

Images include the raw data, fit to the data, residual normalized by peak power "w(1)", and normalized by the individual bin power.

The first pdf includes 0 degrees (normal) and ~5 degrees of tilt for the GC650 (CCD) camera.

The second pdf includes ~10 and ~20 degrees of tilt images for the GC650 (CCD) camera.

The third pdf includes 0 and ~5 degrees of tilt for the GC750 (CMOS) camera.

The fourth pdf includes ~10 and ~20 degrees of tilt for the GC750 (CMOS) camera.

Things to note:
1) GC750 camera seems to have a structure on the camera itself, somewhat circular in nature. One possible explanation is the camera was damage at a previous juncture due to too much light. Need to check earlier images for this problem.
2) GC650 has "bands" which change direction and thickness with angle. Also at higher incidence angle, the sensitivity seems to drop (unlike the GC750 where overall power level seems to stay constant with increasing angle of incidence).
3) GC650 seems to have a higher noise floor,seen from the last plot of each pdf (where each pixel of the residual is normalized by the power in the corresponding pixel of the fit).
  519   Wed Jun 4 16:57:12 2008 josephbConfigurationCamerasDark images from cameras (electronics noise measurement)
The attached pdfs are 1 second and 1 millisecond long integrations from the GC650 and GC750 cameras with a cap in place - i.e. no light.

They include the mean and standard deviation values.

The single bright pixel in the 1 second long exposure image for the GC650 seems to be a real effect. Multiple images taken show the same bright pixel (although with slightly varying amplitudes).

The last pdf is a zoom in on the z-axis of the first pdf (i.e. GC650 /w 1 sec exposure time).

I'm not really sure what to make of the mean remaining virtually fixed for the different integration times for both cameras. I guess 0 is simply offset, but doesn't result in any runaway integrations in general. Although there are certainly some stronger pixels in the long exposures when compared to the short exposures.

Its interesting to note the standard deviation actually drops from the long exposure to the short exposure, possibly influenced by certain pixels which seem to grow with time.

The one with the least variation from its "zero" was the 1 millisecond GC750 dark image.
  520   Thu Jun 5 10:46:26 2008 josephbConfigurationCamerasApproximately uniform reflected white light
In an attempt to investigate the structures seen in previous images for the GC750, I aimed it at a relatively clean section of gray table top roughly a cm or two from the surface and took images (without a lens). As I was holding this with my hand, the angle wasn't completely even with the table, and thus there's a gradient of light in the pictures. However, one should in principle be able to pick out features (such as a circular spot with less sensitivity), but these do not show up.

In my mind, these images seem to indicate the electronics are fine, and suggest that the CMOS or CCD detectors themselves are undamaged (at least in regards to white light, as opposed to 1064nm). An issue with the plastic cap (protective piece) may be the culprit, or perhaps a tiny bit of dust, which the incoherent light from all angles goes around efficiently?

Will try blowing the cameras with clean nitrogen today and see if that removes or changes the circular structure we have seen.
  521   Thu Jun 5 13:35:23 2008 josephbConfigurationCamerasGC750 looking at 1064nm scattered light
I've taken 200 images of the GC750 (CMOS) camera while holding it by hand up to a beam card (also held by hand) in the path of ~5mW of beam power. I then averaged the images to produce the fourth attached plot.

Rob has pointed out the image looks a lot like PCB traces. So perhaps we're seeing the electronics behind the CMOS sensor?

I repeated the same experiment with HeNe laser light (again scattered off a card). These show none of the detailed structure (just what looks to be a large reflection from the card moving around depending on how steady my hand was). These are the first 3 attached plots. So only 1064nm light so far sees these features.

As a possible solution, I did a quick and dirty calibration by dividing a previous PSL output beam by the 1064 average scatter light values. These produce the last attached pdf (with multiple images). The original uncalibrated image is on top, while the very simply calibrated image is on the bottom of each plot.

It seems as the effect may be power dependent (which could still be calibrated properly, but would take a bit more effort than simply dividing), as determined by looking at the edges of the calibrated plot.
  525   Fri Jun 6 16:47:04 2008 josephbConfigurationCameras GC650 scatter images of 1064nm light
Took images similar to the scattered light images from earlier, except with the CCD GC650 camera. The first three attached plots are an average of all 200 images, an average of the first 100 and then an average of the last 100 images.

They show no definite structure. The big red blob which changes with time may be a brighter reflection, although it virtually the same type of setup as the GC750 images.

To do this properly, I should grab a short focal length lens and simply blow up the beam to a size greater than the detector area and simply fix both cameras looking into.

The last set of plots are mean and standard deviation plots from a previous set of runs on 5/29/08 with the GC750 and GC650 running at the same time. The GC650 was receiving approximately 33% of the total power and GC750 was receiving 66% (in otherwords a factor of 2 more).
  530   Wed Jun 11 15:30:55 2008 josephbConfigurationCamerasGC1280
The trial use GC1280 has arrived. This is a higher resolution CMOS camera (similar to the GC750). Other than higher resolution, it has a piece of glass covering and protecting the sensor as opposed to a plastic piece as used in the GC750. This may explain the reduced sensitivity to 1064nm light that the camera seems to exhibit. For example, the image averages presented here required a 60,000 microsecond exposure time, compared to 1000-3000 microseconds for similar images from the GC750. This is an inexact comparison, and the actual sensitivity difference will be determined once we have identical beams on both cameras.

The attached pdfs (same image, different angles of view) are from 200 averaged images looking at 1064nm laser light scattering from a piece of paper. The important thing to note is there doesn't seem to be any definite structure, as was seen in the GC750 scatter images.

One possibility is that too much power is reaching the CMOS detector, penetrating, and then reflecting back to the back side of the detector. Lower power and higher exposure times may avoid this problem, and the glass of the GC1280 is simply cutting down on the amount passing through.

This theory will be tested either this evening or tomorrow morning, by reducing the power on the GC750 to the point at which it needs to be exposed for 60,000 microseconds to get a decent image.

The other possibility is that the GC750 was damaged at some point by too much incident power, although its unclear what kind of failure mode would generate the images we have seen recently from the GC750.
  558   Tue Jun 24 17:12:10 2008 josephb, EricConfigurationCamerasGC750 setup, 1X4 Hub connected, ETMX images
The GC750 camera has been setup to look at ETMX. In addition, the new 1X4 rack mounted switch ( has been connected via new cat6 cable to the control room hub (, thanks to Eric. The camera is now plugged into 1X4 rack switch and now has a gigabit connection to the control room computers as well as Mafalda (

By using ssh -X mafalda or ssh -X, then typing:

cd Prosilica/bin-pc/x86/

A viewer will be brought up. By clicking on the 3rd icon from the left (looks like an eye) will bring up a live view.

Closing the view, and then cd ../../40mCode, and then running ./Snap --help will tell you how to use a simple code for taking .tiff images as well as setting things such as exposure length and size of image (in pixels) to send.

When the interferometer was set to an X-arm only configuration, we took two series of 200 images each, with two different exposure lengths.

Attached are three pdf images. The first is just a black and white single image, the second is an average of 100 images, and the third is the standard deviation of the 100 images.
  566   Wed Jun 25 12:25:28 2008 EricSummaryCameras2D Gaussian Fitting Code
I initially wrote a script in MATLAB that takes pictures of the laser beam's profile and fits them to a two dimensional gaussian in order to determine the position and width of the beam. This code is now (mostly) ported to C so that it can be imbedded in the camera software package that Joe is writing. The fitting works fairly well for pictures with the beam directly incident on the camera, and less well for pictures of scatter off the end mirrors of the arms, since scatter from defects in the mirror have intensities much greater than the intensity of the beam's gaussian profile.

The next steps are to finish up porting the fitting code to C, and then modify it so it can better handle the images off the end mirror. Some thoughts on how to do this are to use a fourier transform and a low pass filter, or to simply use a center-of-mass calculation (with the defect peaks reduced in intensity), since position is more important than beam width in this calculation. The eventual goal is to include the edge of the optic in the picture and use the fit of the beam position in comparison to the optic's position to find the beam's location on the mirror.
  622   Wed Jul 2 10:35:02 2008 EricSummaryCamerasGeneral Summary
I finished up the 2D Gaussian fitting code, and, along with Joe, integrated into the Snap software so that it automatically does a fit to every 100th image. While the fitting works, it is too slow for use in any feedback to the servos. I put together a center of mass calculation to use instead that is somewhat less accurate but much faster (almost instantaneous versus 5-10 seconds). This has yet to be added to the Snap software, but doing so would not be difficult.

I put together a different fitting function for fitting the multiple lorentzian resonance peaks in a power spectrum that would result from sweeping the length of any of the mode cleaners. This simply doesn't work. I tested it on some of Josh Weiner's data collected on the OMC last year, and the data fits poorly. Attempting to fit it all at once requires fitting 80000 data points with 37 free parameters (12 peaks at 3 parameters per peak and 1 offset parameter), which cannot be done in any reasonable time period. Attempting to fit to one specific peak doesn't work due to the corruption of the other nearby peaks, even though they are comparatively small. The fit places the offset incorrectly if given the opportunity (green line in attemptedSinglePeakFitWithoutOffset.tiff and attemptedSinglePeakFitWithoutOffsetZoomed.tiff). Removing this as a parameter causes the fit to do a much better job (red line in these two graphs). The fit still places the peak 0.01 to the right of the actual peak, which worse than could simply be obtained by looking at the maximum point value. Additionally, this slight shift means that attempting to subtract out the peak so that the other peaks are accessible doesn't work -- the peaks are so steep that the error of 0.01 is enough to cause significant problems (red in attemptedPeakSubtraction.tiff is the attempted subtraction). Part of the problem is that the peaks are far from perfect lorentzians, as seen by cropping to any particular peak (OMCSweepSinglePeak.tiff ). This might be corrected in part by correcting for the conversion from PZT voltage to position, which isn't perfectly linear; though I doubt it would remove all the irregularities. At the moment, the best approach seems to be simply using a center of mass calculation cropped to the particular peak, though I have yet to try this.

Changing Josh's code to work for the digital cameras and the PMC or MC shouldn't be difficult. Changing to the MC or PMC should simply involve changing the EPICs tags for the OMC photodiodes and PZTs to those of the PMC or MC. Making the code work for the digital cameras should be as simple as redirecting the call to the framegrabber software to the Snap software.
  657   Thu Jul 10 23:27:57 2008 JohnMetaphysicsCamerasSecret handshakes
Rob and I have joined the ranks of the illuminati and exercised our power.

Osamu showed me the secret way to change the video labels for the quads and
so we fixed them. He made me swear not to divulge this art.

- Rana Adhikari
ELOG V3.1.3-