As discussed at today's run meeting on Teamspeak on the JRPC channel:
The LDAS (DCS) stream of 4 s hoft generation using gstlal_compute_strain started writing to LDAS disks at 1125451520 == Sep 05 2015 01:25:03 UTC.
There already exist two redundant DMT streams of 4 s hoft writing to LDAS and DMT (GDS) disks. These are now vetted for diffs in the STRAIN and ODC channels.
Aggregation of hoft into 4096 s frame files for offline analysis is now configured (via the diskcache) to use the DMT hoft if the STRAIN and ODC channels agree in the two DMT streams; otherwise it will use the LDAS stream. (This can be easily switched back to using only the two DMT streams as it has been using prior to the time given in this alog, if problems arise).
Injection |
Time of first injection, UTC |
Injection spacing |
Total number of injections |
Good channels for environmental signal |
|
Sept. 5 |
|
|
|
crowd (10 people) walking randomly for 2 or 3s periods in control room |
1:39:00 |
5s |
12 |
HAM2 seismometer, vertex seismometer H1:PEM-CS_ACC_LVEAFLOOR_HAM1_Z
|
crowd (10 people) walking randomly in hall just outside control room |
1:41:00 |
5s |
12 |
same |
External door near control room shutting by itself |
1:44:00 |
5s |
12 |
same |
chair rolling in control room (most common chair) |
1:46:00 |
5s |
12 |
same |
Single large steps in control room |
1:50:00 |
5s |
12 |
same |
Airlock door (door to lab area near control room) shut by hand |
1:52:00 |
5s |
12 |
same |
slamming office door (Sheila, Kiwamu, Me) |
1:54:00 |
5s |
12 |
same |
hammer dropped from waist height in vacuum lab |
1:59:00 |
5s |
12 |
same |
setting car battery down in OSB shipping area |
2:02:00 |
5s |
12 |
same |
continuous bouncing for 5s on exercise/seating ball |
2:07:00 |
10s |
6 |
same |
C. Cahillane I have been busy trying to calculate kappa_tst, kappa_pu, kappa_C,and f_c from ER8 data so I might begin finding the uncertainty associated with these numbers. The uncertainty in all of these calibration coefficients depend heavily on our calibration line measurements (See T1500377 for the calculation of these coefficients from model params and calibration lines). I was curious to see if uncertainty in our calibration line measurements will be a significant source of uncertainty in the total budget, or if I can ignore it for now. I took 100 GPS times starting at 1125316818, each ten seconds apart, and computed the 10 second FFT of DARM_ERR, X_TST, X_CTRL, and X_PCAL. Then, I found their values at the calibration line frequencies at H1: f_tst = 35.9 Hz f_pcal = 36.7 Hz f_ctrl = 37.3 Hz f_pcal2= 331.9 Hz and took the following ratios: X_TST(f_tst) / DARM_ERR(f_tst) X_PCAL(f_pcal) / DARM_ERR(f_pcal) X_CTRL(f_ctrl) / DARM_ERR(f_ctrl) X_PCAL(f_pcal2) / DARM_ERR(f_pcal2) and plotted their amplitudes below. Percent uncertainties: X_TST(f_tst) / DARM_ERR(f_tst) = 0.98% X_PCAL(f_pcal) / DARM_ERR(f_pcal) = 0.95% X_CTRL(f_ctrl) / DARM_ERR(f_ctrl) = 0.84% X_PCAL(f_pcal2) / DARM_ERR(f_pcal2) = 1.05% Since I am just starting to compute the uncertainty expressions for the calibration coefficients in Mathematica, this study informed me that there is ~1% uncertainty in all of our calibration line amplitudes, which is significant enough to be included in all uncertainty calculations. Two notes: (1) I am assuming for now there is no quantization noise in our digital signals DARM_ERR, DARM_CTRL, X_TST, X_CTRL, and X_PCAL. This is almost certainty a secondary consideration in the total budget for now. (2) The method I used for dewhitening X_PCAL is known to be incorrect, so their absolute values should not be taken too seriously. But since all I cared about here is the statistical uncertainty, any systematic errors are a simple gain that is common to all data points, so this result is still valid.
C. Cahillane, E. Hall I have made some updates because of Evan. He suggested that I switch the measurements to be Response / Excitation as is the norm with transfer functions, so now my plots are of DARM_ERR / X_{TST, CTRL, PCAL}. In addition he suggested that I take more data and make histograms of the DARM_ERR / X_{TST, CTRL, PCAL} plots to better see the distribution, to see if it looks like a Gaussian or Rayleigh distribution. (Plot 2 is the Histogram, Plot 1 is the standard scatterplot) Plot 3 is the line amplitude. In the case of PCAL, we can make readouts, leading to the statistical uncertainty we see only in X_PCAL on this plot. The others, X_CTRL and X_TST, we cannot make readouts and must trust that our excitation is constant over all time. Plot 2 looks like a nice Gaussian distribution for all the observed calibration lines. We will still need to consider these statistical uncertainties in the total uncertainty budget.
While Jeff and Darkhan have been trying to get the actuactor coefficients right for calibration, I worked on an orthogonal task which is to check out the latest optical gain of DARM.
Summary points are:
The plots below show the measured optical gain measured by Pcal Y with the loop suppression taken out by measuring the DARM supression within the same lock stretch.
I used the data from Aug 28 and 29th (alog 21190 and alog 21023 respectively). The parameters were estimated by the fitting function of LISO. I have limited the frequency range of the fitting to be avove 30 Hz because the measurement does not seem to obey physics. I will metion this in the next paragraph. The cavity pole was at around 330 Hz which claims a bit lower frequency than what Evan indendently estimated from the nominal Pcal lines (alog 21210). Not sure why at this point.
One thing we have to pay attentin is a peculiar behavior of the magnitude at low frequencies -- they tend to respond lesser by 20-30 % at most while the phase does not show any evidence of extra poles or zeros. I think that this behavior has been consistently seen since ER7. For example, several DARM open loops from ER7 show very similar behavior (see open loop plots from alog 18769). Also, a recent DARM open loop measurement (see the plot from alog 20819). Keita suggested makeing another DARM open loop measurement with a smaller amplitude, for example by a factor of two at a cost of longer integation time in order to detemine whethre if this is associated with some kind of undesired nonlinearity, saturation or some sort.
I did a similar fit that Shivaraj did at LLO (alog # 20146), to determine the time delay between PCAL RX and and the DARM_ERR. Both signal chain have one each of IOP (65 KHz), USER model (16 Khz) and AA filter between them. The expected time delay between the PCAL and DARM_ERR as shown in the diagram below should be about 13.2 us in total. I used the Optical gain as 1.16e+6 from the alog above and fitted for cavity pole and time delay. I got cavity pole estimate of 324 Hz, close to what Kimamu got from his fitting and time delay of 21 us. This is 7.8 us more than what we expected from the model.
9/4 DAY Shift: 15:00-23:00UTC (08:00-16:00PDT), all times posted in UTC
Summary: Able to get a few hours of Observation time during the shift. Also had a few hours of PEM Injections by Robert. Coordinated a bit with LLO with regards to when to drop out of Observation as well. All of this is pointing us toward a goal of having long double-coincidence duty cycles for the next week so we can get an idea of the state of our machines going into O1.
Robert plans to continue with injections into the evening so he can have a lighter day tomorrow.
Handed off a nice H1 (~70Mpc) to Nutsinee with quiet seismic and slight winds.
Poll of Control Room Work: PCal/OMC model work, Calibration Analysis, PEM Injection analysis, Ops Script work.
Support: Had Commissioners around, but not needed since H1 was locked the whole time (since 16UTC / 10pm PST)
Day's Activities
ECR: https://dcc.ligo.org/LIGO-E1500373
userapps/cal/common/models/PCAL_MASTER.mdl
userapps/omc/h1/models/omc.mdl
userapps/omc/common/models/omc.mdl
h1calex, h1caley and h1omc were all successfully built but not installed. These will be installed on next Tuesday.
The seized up compressor on the supplemental chiller unit for the staging building, commonly described as the AAON Unit has been replaced, charged, and is now running at 100%.
Chris S. Joe D. Both 70% This week the guys have cleaned the original caulking and installed metal strips on 200 meters of tube enclosure on the top side. They have also cleaned the caulking on an additional 60 meters of enclosure.
I had stepped out the control room for ~15min, and when I came back the first thing I noticed was a YELLOW "OK" on the GWIstat screen. I asked around the Control Room to see if anyone knew why we were out of Science Mode (granted I should have announced my exit), but no one seemed to notice the drop from OBSERVATION.
I took us back to Observation Mode immediately since there was no reason given for having us out. I looked in the Verbal Alarm terminal, and did not see a note of the Intent Bit being changed. (So, I've talked to TJ about getting this in the Verbal Alarm script. And to also have time stamps attached to the Intention Bit alarms so we'll know when these come up.
Keita admitted he was the culprit; he opened a DTT session [which has an excitation], and then started it. (We tested this, while in Observation mode, and we were able to open this DTT session, which had excitation selected, and this did not drop us out. Robert & I thought that just opening a session would drop us out. He mentioned that maybe this is the case with AWG. We should test whether it can drop us out.). So this is when we were out:
16:19-16:29UTC Out of OBSERVATION
Addendum:
Prior to PEM Injections, Robert and I wanted to check if AWG does in fact have a different effect with regards to staying in Observation Mode with Excitation channels. Robert opened an AWG session (no drop), but once he merely selected an excitation in AWG, we were dropped out of Observation Mode. This is even WITHOUT hitting the "Set/Run" button!
When you do this, the DIAG_EXC guardian Node gets an orange box and has the message: EXC: [system] excitation!
We were dropped out of Observation at: 20:10UTC due to this excitation being selected (and NOT run) (this was different from what DTT did).
The noise below 30 Hz looks a bit non stationary, see the zoom in of the attached spectrogram.
The behavior of the noise reminds me very much of scattered light, but I'm not 100% positive right now.
For some reason Evan is reluctant to make a new post about it: LHO alog 21210, comment 5 (H1NB_2015-08-27_123000.pdf)
Due to a crazy big offset of -0.5 in Y_TR_B_PIT (for SOFT modes sensing), Y IR QPDB is almost always railing a bit in 24W operation, and Y IR QPDA is not too far.
Next time IFO drops out of lock, somebody needs to lower the whitening gain by 3dB and set a new dark offset for each quadrant.
These whitening gains are controlled by the ISC_LOCK guardian. We already lower them by 6 dB in the DRMI_ON_POP state (which produces the momentary fake jump in arm power that everyone asks about), so it sounds like we should be lowering them by 9 dB instead.
[Shamefully, we don't change the dark offsets when we change the whitening in this step.]
Shame.
DRMI_ON_POP now turns down whitening gain from 18 dB to 9 dB.
The landscapers will be out this weekend-Saturday and possibly Sunday for weed control in front of the OSB, Staging Building, and the LSB.
Patrick handed off an H1 which has been at Nominal Low Noise since about 5:30utc (10:30pmPST) with a range hovering around 70Mpc. There was a noticeable step down in range (~60Mpc) at 12:04UTC for about 30-60min (Patrick notes several ETMy saturations around 12:10UTC & Ed notes there were earthquakes aroudn 12:15UTC).
Injection around 11:15UTC?
Looks like we have been in Observation Mode since 6:00UTC, but GWIstat says we've only been in this state since about 11:15UTC (this isn't Peter's Burst from last night....maybe a Transient Injection? Or some other injection? Is there a schedule for these things??).
Today's Outlook: PEM Injections for good chunk of day
Will stay in Observation a little, but Robert says he said would like to start PEM Injections within an hour or so (~16:45UTC?), and will be PEM Injecting for a good chunk of the day so he does not have to do much on Saturday. Have just talked with Lisa at LLO and they will be interested when we go out of Observation, so they could also go out and do some much-need commissioning.
Seismically, we look fairly quiet with microseism noticeably trending down by 0.1um/s over the last 24hrs. Winds are also quiet.
ER8 Day 18, no restarts reported.
Dan, Daniel, Evan
The addition of a 9.1 MHz bandpass on the OCXO output has removed the broadband excess noise between DCPD sum and null. The dashed lines in the figure show the sum and the null as they were three days ago (2015-08-31 7:00:00 Z), while the solid lines show the sum and the null after the filter was inserted.
Since at least June (probably longer), we've had a broadband excess noise between the sum and null DCPD streams. Stefan et al. identified this as 45.5 MHz oscillator noise a few weeks ago (20182).
In parallel, we switched the 9 MHz generation from an IFR to the OCXO (19648), and we installed Daniel's RFAM driver / active suppression circuit (20392), but the excess noise remained (20403). For a while we suspected that this was 45.5 MHz phase noise (and hence not supressed by the RFAM stabilization), but the shape and magnitude of the oscillator phase noise coupling (20783) were not enough to explain the observed noise in the DCPDs, under the assumption that the OCXO phase noise is flat at high frequencies (20582). For that matter, the shape and magnitude of the oscillator amplitude noise coupling were also not enough to explain the observed noise in the DCPDs, assuming a linear coupling from the RFAM (as sensed by the EOM driver's OOL detector) (20559).
Daniel et al. looked at the 45.5 MHz spectrum directly out of the harmonic generator in CER, and found that most of the noise is actually offset from the 45.5 MHz carrier by 1 MHz or so (20930), which is above the bandwidth of the RFAM suppression circuit. This suggested that the noise we were seeing in the DCPDs could be downconvered from several megahertz into the audio band.
Yesterday there was a flurry of work by Keita, Fil, Rich, et al. to find the source of this excess noise on the 45.5 MHz (21094 et seq.). Eventually we found circumstantial evidence that this excess noise was caused by baseband noise out of the 9.1 MHz OCXO.
Tonight we installed a 9.1 MHz bandpass filter on the OCXO output. This has removed the huge 1 MHz sidebands on the 45.5 MHz signal, and it also seems to have greatly lowered the coherence between DCPD A and DCPD B above a few hundred hertz.
The chain from OCXO to filter to distribution amplifier currently involves some BNC, since we could not find the right combination of threaded connectors to connect the filter to the amplifier. This should be rectified.
Also, it appears that our sum is lower than our null in a few places (400 Hz in particular), which deserves some investigation.
"NULL>SUM" problem is just DARM loop. You're talking about 10%-ish difference, and DARM OLTF gain is still 0.1-0.2 at 400Hz.
See attached.
I don't know how to obtain official DARM OLTF model, so I just took 2015-08-29_H1_DARM_OLGTF_7to1200Hz_tf.txt in
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/ER8/H1/Measurements/DARMOLGTFs/
The coherence for this OLTF measurement was much larger than 0.95 for the entire frequency range shown on the plot.
On the bottom is |1+OLTF|. I interpolated this to the frequency spacing of SUM and NULL spectra, and plotted SUM*|1+OLTF|, SUM, and NULL at the top.
Note that DARM OLTF template measures -1*OLTF.
Nice work. After O1 we can figure things out now you have narrowed it down.
Nice work!
Great job! Following up on the discussion during the commissioning meeting today, at LLO Evan's equivalent plot of the coherence between the two OMC PDs is already below 10^-3 (below 3 kHz).
Fil and I replaced the BNC cable with an SMA/N cable.
This entry is meant to survey the sensing noises of the OMC DCPDs before the EOM driver swap. However, other than the 45 MHz RFAM coupling, we have no reason to expect the couplings to change dramatically after the swap.
The DCPD sum and null data (and ISS intensity noise data) were collected from an undisturbed lock stretch on 2015-07-31.
Noise terms as follows:
The downward slope in the null at high frequencies is almost certainly some imperfect inversion of the AA filter, the uncompensated premap poles, or the downsampling filter.
* What is the reasoning behind the updated suspension thermal noise plot?
* Its weird that cHard doesn't show up. At LLO, cHard is the dominant noise from 10-15 Hz. Its coupling is 10x less than dHard, but its sensing noise is a lot worse.
I remade this plot for a more recent spectrum. This includes the new EOM driver, a second stage of whitening, and dc-lowpassing on the ISS outer loop PDs.
This time I also included some displacement noises; namely, the couplings from the PRCL, MICH, and SRCL controls. Somewhat surprising is that the PRCL control noise seems to be close to the total DCPD noise from 10 to 20 Hz. [I vaguely recall that the Wipfian noise budget predicted an unexpectedly high PRCL coupling at one point, but I cannot find an alog entry supporting this.]
Here is the above plot referred to test mass displacement, along with some of our usual anticipated displacement noises. Evidently the budgeting doesn't really add up below 100 Hz, but there are still some more displacement noises that need to be added (ASC, gas, BS DAC, etc.).
Since we weren't actually in the lowest-noise quad PUM state for this measurement, the DAC noise from the PUM is higher than what is shown in the plot above.
If the updated buget (attached) is right, this means that actually there are low-frequency gains to be had from 20 to 70 Hz. There is still evidently some excess from 50 to 200 Hz.
Here is a budget for a more recent lock, with the PUM drivers in the low-noise state. The control noise couplings (PRCL, MICH, SRCL, dHard) were all remeasured for this lock configuration.
As for other ASC loops, there is some contribution from the BS loops around 30 Hz (not included in this budget). I have also looked at cHard, but I have to drive more than 100 times above the quiescient control noise in order to even begin to see anything in the DARM spectrum, so these loops do not seem to contribute in a significant way.
Also included is a plot of sensing noises (and some displacement noises from LSC) in the OMC DCPDs, along with the sum/null residual. At high frequencies, the residual seems to approach the projected 45 MHz oscillator noise (except for the high-frequency excess, which we've seen before seems to be coherent with REFL9).
Evidently there is a bit of explaining to do in the bucket...
Some corrections/modifications/additions to the above:
Of course, the budgeted noises don't at all add up from 20 Hz to 200 Hz, so we are missing something big. Next we want to look at upconversion and jitter noises, as well as control noise from other ASC loops.
I've attached the result. Quick observation from the spectrograms tell me that dropping hammer from waist height in vacuum lab (aka Kyle's lab) and setting car battery down in OSB shipping area coupled into DARM.
Instead of plotting the spectogram starting at the time of injection, this time I did +/- 10 seconds instead so it's more clear when the injection started and when it's ended.
Since I have never really explained what I did in finding noises in the sensors and in DARM, I'm taking some time to explain it in this alog. First I looked for injections in appropreiate sensors (accelerometer, microphone, seismometer). I zoomed in the spectrograms until I get the signal to show up >3 pixels or so in both frequency and time domain (as Robert suggested). Once I was certain that the injections were there, I went to look for them in DARM. I started with the same frequency range that the injections showed up in the sensors I used, if I couldn't find anything then I move up/down in frequency domain to look for any possible up/down conversions with increments depending on how wide the injections were in terms of frequency. Until I reached the lowest/highest frequency that the injections could possibly showed up, if I still didn't see anything only then I would conclude that the injections didn't show up in DARM.
More investigations on how some of these injections coupled into DARM will take days, weeks, I don't know. The point of these injections was to determine what activities can and cannot be performed during Observing run. This is why I posted spectrograms first and not wait until I'm done with the analyses. I simply let DARM spectrogram speaks for itself.
I have also attached better spectrograms of the super ball injection, produced using Dan's script.