See the attched figure.
The violin modes are not harmonics of each other, as we know. But they aren't harmonic in a rather interesting way, because the frequencies are less than harmonic. We might expect that the mode frequencies be greater than the harmonic values because of the bending stiffness of the fiber. Only a completely floppy string is harmonic. The fiber's modulus of elasticity will tend to become more important at higher mode numbers, increasing the frequencies.
However, we observe that the modes are lower in frequency than the harmonic values. The attached plot shows how close (or far) the modes are to harmonic. The y axis is the ratio of a mode's frequency to the first mode. The x axis is the mode number. An ideal harmonic system has a slope of 1. This is represented by the dashed black line. The blue line is from an FEA done by Alan Cumming at Glasgow University. This FEA takes into account the non-uniform shape of the fiber. The green line is from measurements of LHO ETMY.
The FEA does indeed predict that the fiber should be less than harmonic. This could be because the 400 micron diameter fiber has 20 mm long 800 micron sections at the ends. Perhaps the effective bending point moves along the 800 micron section.
However, the measurements are even less harmonic than the FEA. It is not clear why this is. One possibility, though this sounds a bit crazy (but I can't think of anything else), is that the fiber is pitching the PUM and test mass as it oscillates. If the masses pitch, it makes the fiber behave as if it were longer, moving the frequencies down. Higher modes might exert more pitch torque becasue the fiber has more slope at its ends.
Running some numbers on the fiber length, The nominal value of 587 mm gives a fundamental frequency of 507 Hz, very close to ETMY (508 Hz) and within the spread of the other suspensions. But for ETMY mode 4, I have to increase the length by 20 mm. For ETMY mode 7 this is nearly 40 mm, which puts the bending point somewhere in the ears I think. For reference, the centers of mass are 602 mm apart.
The mode frequencies in Hz used in the attached plot are
FEA from Alan Cumming:
511, 1017, 1519, 2013, 2515, 3020, 3470
LHO ETMY:
508.3, 1009.1, 1484.5, 1956.7, 2425.8, 2880.6, 3333.2
The LHO ETMY data came from the following alogs
17610 - mode 1, averaged over the 4 given frequencies
17365 - modes 2 and 3. Mode 2 is averaged over the 4 given values. Mode 3 only has 1 value.
18764 - mode 4. This list doesn't specify which modes are which suspension. I chose to average the 4 highest values assuming they are ETMY because for all other modes ETMY is the highest.
18614 - modes 5, 6, and 7. I averaged the 2 given values for each of these modes.
Evidence for fiber stiffness being a function of frequency? Is the fiber stiffness in the FEA consistent with Kramers-Kroenig and the anelasticity assumed for the losses?
For what it's worth, we have also observed (and damped) an eighth harmonic of ETMY - 3800.995 Hz.
I'm not exactly sure how the FEA handles the stiffness. Without reading all the details, I know there is some description of it in "Design and development of the advanced LIGO monolithic fused silica suspension", Cumming, et al.
I added the 8th ETMY mode to the plot and reattached it here. Thanks Dan.
Alan Cumming pointed out that the FEA does not match the anharmonicty because the FEA here is for the MIT quad suspension. FEA data on slide 5 of is G1700038-v1 shows a quite good match.
I followed up the range drop tonight at 2015/06/07 around 8:30 UTC. Here are the symptoms: - 25.4Hz and harmonics ringing up and saturating everything. - Significant increase of two lines: 842.781Hz and 868.188Hz (Yes, they are 25.407Hz apart) Those are marked with crosses in the attached plots. Note that there are some other lines that increased (red) over the reference (black), but they are symmetric 25.4Hz sidebands of strong lines that did not increase. The two lines at 42.781Hz and 868.188Hz are not such modulation sidebands. Parametric Instability? alog 17903 reports on an observed parametric instability at 15540.6Hz, causing a line at 843.4Hz. Seems close enough to suspect it as the culprit for 842.781Hz. But what is 868.188Hz? And why does the 25.4Hz show up so strong? BS roll mode? My first though on seeing something at 25Hz was BS roll. But T1200415 reports the BS roll mode at 25.9715Hz... If we believe that, 25.4Hz can't be the BS roll... Do we have an actual recent measurement of the BS roll? I did try to look at the BS oplevs for a sign of the roll mode rung up - nothing. I haven't looked at the OSEMS yet All attached plots were taken at 8:15 UTC, just before it go really bad. The black reference is from 7:00 UTC. I also left instructions with Cheryl on how to lower power if this happens again. If that fixes it it would nail the PI. Running of to the airport, but Cheryl will follow up.
If you want to look at the trans QPDs you have to look at the IOP channels, is that what you were looking at?
Chatted with Will (LLO Operator) early in the shift. He said he and Carl were working on getting L1 UP (& they did shortly after).
5:08 landscaper truck on site, and they seem to be driving noticeably slow around the site (a good thing!).
For the current lock segment, it's been odd in that the range started out nice at 62Mpc, but over the following hours it trended down and leveled off at 58Mpc. I hand this segment over to Cheryl.
NOTE: Also talked with Sheila again. If the odd 25.4Hz noise comes back she suggested going to lower power (for Operators that would mean going to Manual Mode ith the ISC_Lock Node, selecting "Increase Power" on the Guardian, and then taking the PSL rotation stage to 17W).
BS ISI Watchdog Trip
During this current segment, Ed noticed from home and emailed me about a RED ISI for the BS. Turns out the Stage2 Watchdog for the BS ISI GS13 was tripped (so red blocks for Stage2 FF & ISO....BUT the BS is unique in that we don't use the ISO for Stage2 anyway). So we could do a RESET ALL and this shouldn't affect anything, but we (Cheryl & I) are deciding to do this after the segment ends. [Thanks to HUGH for help over the phone...and Ed for keeping an eye remotely!]
Due to BS's unique configuration for the ISI, we can ride through a WD trip like this. (but it's something which could go unnoticed unless you happen to be watching the OPS overview). For other ISIs, if a WD trip happened on this stage, the ISO filters are in effect and they definitely would raise one's attention because things would definitely (most likely) swing around.
So maybe we should have an audio alarm for this special case? Or maybe we can NOT have the WD affect this path in some way....just a thought.
model restarts logged for Sat 06/Jun/2015
no restarts reported
As soon as L1 locked up, H1 nosedived with a huge noise source at 25.37Hz (big ugly peak with sidebands) & its harmonics (Sheila could not find a SUS resonance at this frequency). This dropped sensitivity from 60Mpc to 30Mpc. After about 15min, I intentionally broke lock to see if a new lock would improve sensitivity. (Note: It wasn't clear to me how to break lock with our new fandangled aLIGO H1 detector. I hit DOWN in the ISC Lock Guardian Node, but this just made the whole node fault RED. Ultimately, I went to the LSC Overview & pushed the OFF button for LSC Mode & that did the trick.
I wanted to quickly go back up, so I saw the Yarm was a bit ugly, so I tweaked on ETMy, but could only get up to 0.86counts (I thought WFS would kick in to get me over 1.0). ISC_LockGuardian spouted off about "waiting for arms to lock" and never progressed with locking. At this point, I went to the phone, and got a hold of Sheila. She mentioned I could tweak TMSy, but I might do better to do a full Initial Alignment. I ended up trying both.
Alignment was fairly straightforward (this was really only my 4-5th time to go through and alignment, AND I've never sat at the chair and had H1 lock by my own powers!).
Then I went for locking. First attempt, I waited about 3-4min for DRMI to lock, but it dropped out during the "Switch to qpds". Second attempt had me wait a whopping 3min for DRMI to lock. And for this, lock I went all the way up to LSC_FF (Bounce modes were nice and well below 10^-12).
Once all the way through, I wasn't completely clear on what I had to do with the ETMx ESD Driver. ESD ACTIVE was RED, but the monitors were all a few hundred counts negative (range was about 55Mpc). So, I clicked on the "HI" button, which made ESD ACTIVE go GREEN briefly and then back to RED, and this sent the monitors close to zero (and range went up to 62Mpc).
SDF looked all GREEN, and Intent Bit set to Undisturbed at 3:58am (10:58utc), and H1 joined L1.
[Time for lunch.]
Sheila suggested I take a look at the TMS QPDs to look for the noise source observed earlier. However when I looked at the TMS QPDs, I did not see any of the noise. However I was able to see the 25.4Hz noise on the OMC QPDs.
I am having trouble saving pdfs from DTT so I just took a snapshot of my DTT session (the references are from the noisy period). Also including the Cal Delta L channel (i.e. DARM) which shows the harmonics of this noise source.
Not much to report here. I walked in with Cheryl and Stefan locking H1 and then I didn't touch it at all.
John called and said that he saw the EY VEA temp starting to rise so he turned off a heat source and it leveled back out nicely. It went up to a max of about 68.6F
There was one BIG glitch at ~20:30 pst but it held lock.
I have to say how impressed I am with the lock stretches lately, a lot of hard how is really starting to show.
Handing it off to Corey, good luck!
The summary pages report 18h of lock in low noise with an average of 65 Mpc (WOW!). I believe the longest lock of the aLIGO era is the 30h long lock of L1 (no pressure! :-) However, I think this H1 lock now sets the new record of the best time-volume lock ever, equivalent to about ~26 days of eLIGO @ 20 Mpc.
LIKE! I have the remote MEDM screens. I've been stalking. :) The only bummer is, it looks like coincidental locking time isn't quite as good.
The YEND is having trouble keeping cool in the hot part of the day. We can make adjustments on maintenance day or sooner if needed.
TJ and I talked on the phone and decided that I should proceed and turn off the remaining heat in the VEA. We will monitor the response tonight. Tomorrow or Tuesday Bubba and I need to lower the chilled water setpoint at YEND. For some reason it is 10F higher than Xend. I thought we had made the site rounds and set all chillers for the spring heating season.
IFO was locked most of the shift.
Broke lock 30 minutes before shift end. Relocked to Violin and Bounce Mode, then handed off to Stefan and TJ.
Glitches in the IFO look suspiciously like Big, Smaller, Smaller, Smallest, and at the time I left I could see three sets of glitches that look that way.
I monitored the ASC OUTMONs for all the optics, and IM4 glitched at some point and PRM and SRM did as well, but more like a response.
Attached is my striptool of the ASC OUTMONs at Lock Loss, which also shows that IM4 went first and lock loss followed.
Using two hours of undisturbed data from last night's 66 Mpc lock, I repeated Den's sum/null stream analysis in order to see if we have a similar 1/f1/2 excess in our residual.
I took the OMC sum/null data (calibrated into milliamps), undid the effect of the DARM OLTF in order to get an estimate for the freerunning OMC current, and then scaled by the DARM optical gain (3.5 mA/pm, with a pole at 355 Hz) to get the equivalent freerunning DARM displacement. The residual is then the quadrature difference between the sum and null ASDs.
The attachment shows the sum, null, and residual ASDs, along with the anticipated coating Brownian noise from GWINC. [Just to be clear: the "sum" trace on this plot corresponds to our usual freerunning DARM estimate, although in this case it comes purely from the error signal rather than a combination of the error and control signals.]
If there is some kind of excess 1/f1/2 noise here, it is not yet large enough to dominate the residual. Right now it looks like the residual is at least a factor of 2.2 higher than the expected coating noise at all frequencies. We already know some of this is intensity noise.
The other thing to note here is that we are evidently not completely dominated by shot noise above 1 kHz.
I repeated this on a lock stretch from 2015-06-07 00:00:00Z to 02:00:00Z, but the result is pretty much the same. The best constraint we can put on coating noise right now from the residual is about a factor of 2.2 higher than the GWINC prediction. I also think the residual is not yet clean enough in this frequency band to make an inference about its spectral shape.
I tried increasing the CARM gain by 3 dB to see if the residual would decrease, but it does not (except maybe round 6 kHz; see the attached dtt pdf). So this broadband excess in the sum may not be frequency noise.
There is an error in the above plots.
Only the DCPD sum should be corrected by the DARM OLTF to get the equivalent freerunning noise. The DCPD null should not be corrected. To refer to noise to DARM displacement, however, all these quantities must be corrected by the DARM cavity pole.
This time I've included the DCPD dark noise (sum of A and B), also not corrected by the loop gain.
A few more corrections and additions:
Jess, Andy, TJ, Duncan, Detchar,
In entry 18783 (at 19 UTC on June 2) Jeff et al performed an autocal on the SUS DACs for the Modecleaner, in response to the report in log 18739. He asked Detchar to report if the glitches are gone, if they come back over time, etc.
What we've found so far is that the DAC glitches are still present on MC2 M3 zero crossings. Their amplitude (seen in LSC MCL and IMC alignment channels) has reduced by a factor of 2. And their amplitude does not appear to be increasing over time since the restart, on the timescale of days.
Figure 1 shows normalized spectrograms of the DAC glitches witnessed by LSC MCL before and after the Autocal.
Figure 2 shows a time vs SNR plot of the individual glitches (only during observation intent time) in LSC MCL (at frequency < 200Hz to be dominated by DAC glitches). The many glitches with signal-to-noise ratio of 30 are now many glitches with SNR 15-20. Autocal occurs at hour 19.
Figure 3 shows the same plot for IMC-DOF_1_P_IN1_DQ, another good witness of these glitches.
Figure 4 caption: The Glitchgrams on the bottom show glitchiness of IFO, not related to the MC2 DAC glitches (we think) but just to see when IFO is locked and in good state. The top plots are all normalized spectrograms. The left two plots are before autocal, MCL glitches are really loud. The right five plots are after autocal, most glitches are less loud. Where "loud" is assessed by top of color bar (admittedly poor measure).
Figure 5: Same for two IMC alignement channels DOF 1 P and DOF 2 Y that are both good witnesses of DAC glitches.
Jess wrote nice scripts to make the glitch vs time plots so we will keep an eye on these to see when/if the amplitude increases.
Notes: Sorry Jeff, Peter, et al, I realize now that amplitude rather than SNR, calibrated units, and the SUS VOLTMON channels would be more useful for assessing the size of the DAC glitches. We'll work on that.
We've also made timeseries plots of the glitches in the noisemons before, just after, and several hours after the auocalibration. The glitches seem to come back most strongly in the LL DAC. The first three slides have a 10,100 Hz bandpass so the glitches can be seen clearly. The last two slide have just one second of data, so a single glitch can be seen in the raw data.
Hang, Stefan Our lock is at 14h and counting. Attached are the output control signals for the 10 optics that are controlled by the ASC system. (All of them are relieved to the top mass, so the top mass output is what I am plotting.) There is some funny behaviour in PR2 pitch, and some of it also shows up in the BS pith nd yaw. Also - interestingly, the sharp rise in SR2 PITCH about 10.5h ago (2015/06/06 07:55 UTC) corresponds to a drop in the inspiral range by about 10% (6Mpc), and comes from the 80Hz to 300Hz region. Whatever this noise is, it seems to be related to SRC alignment. The third attached plot shows the difference between before that SR2 move (2015/6/6 7:32 UTC, 67Mpc) and after (2015/6/6 8:32 UTC, 61Mpc). Sounds like SRC ASC work is not done yet.
Here are some plots of the AS_A_RF36 and AS_B_RF36 signals around the time of the SRC cavity shift (2015/06/06 07:55 UTC). PITCH: AS_B_RF36_I and AS_B_RF36_Q are servoed to zero - the don't see any jump. AS_A_RF36_I and AS_A_RF36_Q both show a clear jump at that time, so there is a good chance to find a better PIT error signal. YAW: The jump does show up in the two I sensors in YAW as well. While the current YAW input matri seems to work for stability (18h locks), I am still not sure about its lock-point. More work needed.
GerardoM and RickS GUIDANCE FOR A SYSTEM THAT IS ALREADY LOCKED On the PSL_ISS.adl MEDM screen (see attached image), look at the strip chart in the top-right corner. The diffracted power level should be about 7%. A few percent more or less is OK, but I suggest setting to near 7% at least once per week, say Tuesday during the maintenance period. To change the diffracted light power, one adjusts the “REFSIGNAL” field in the lower left corner. A change in this parameter of 0.01 changes the diffracted power by about 1%, so make small changes. A larger negative number (say going from -2.00 to -2.01) will decrease the diffracted light level. This REFSIGNAL field is the DC laser power level (ignoring the minus sign) that the servo compares with the “Output AC” level on the PD that is selected in the middle-left portion of the screen. Note that in the screen snapshot the REFSIGNAL is at -2.03 and the PD A Output AC signal is at 2.03. This indicates that the loop is operating properly; the loop tries to make the PD output be equal to the reference level (without the minute sign, of course). Notice that the diffracted light level is varying a bit but is close to 7% on the strip chart. At the middle-right edge of the screen the Diffracted Power field indicates 7.38%. This is the field that is plotted in the strip chart. GUIDANCE FOR WHEN THE SYSTEM IS NOT LOCKED In the case that the ISS servo is not locked and you are having difficulty locking it, I suggest the following: With the loop unlocked (Autolock OFF), observe the PD A AC output level. This may be a bit hard to do if the value is swinging a lot quickly. Set the REFSIGNAL level to about ten percent below this observed mean value. Close the loop (Autolock ON) and observe the diffracted light time series in the strip chart. If the diffracted light level increases and goes off screen at the top, then your REFSIGNAL setting is too low (absolute value is too small) so you are not requesting enough light and the servo is trying to diffract a lot of light to give you the low level you requested. Increase the (absolute value) of the REFSIGNAL field. If the diffracted light level decreases and goes off screen at the bottom, then your REFSIGNAL setting is too high (absolute value too large) and you are requesting more light than the servo can give you and still maintain some diffracted light headroom. Decrease the (absolute value) of the REFSIGNAL field. Once the system stabilizes, set the diffracted light level to be close to 7% by making small adjustments to the REFSIGNAL value. Be patient, the time constant is pretty long and small changes make a big difference (on order one or two percent per 0.01 increments in the REFSIGNAL value). Once the diffracted light level is near 7%, observe a few minutes of the strip chart data. The variations should be on the order of what is shown in the attached screenshot. If all else fails, feel free to call me (Rick) at any hour, any day, and I will try to help over the phone. My numbers are in the site directory.
The reference to PD A only applies to the image provided. We are currently using PD B as the in loop PD. In either case, the graphic provided on the medm screen will show the path of the loop.