Detector Watch S6 V3

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458
Topic 194728

This continues from this thread. Catching up the logs. Hanford :

- alot of nearby activity in & around the old nuclear enrichment site. Here is a local news service article on that. It seems there might be about 1kg or so of Plutonium about. If that doesn't sound like alot to you then read this. It truly is the devil's own substance - hence the name.

- several mentions of LUMIN which is a trigger from the GW detectors to 'ordinary' telescopes.

- this is what a passing aeroplane 'sounds like' when the IFO picks it up :

Note the Dopplering over 160Hz ( approaching ) down to 60Hz ( receding ) over about 3 minutes. I wonder if that's ( an harmonic of ) a helicopter rotor blade?

- some various investigations/correlations in the noise budget.

Gotta fly, be back.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

Detector Watch S6 V3

Sorry for the pause, it has been the silly season DownUnda! :-)

Now for Livingston ( since 06/01/10 ) :

which is a MATLAB graphic showing the fit between a model in blue and data in red, the difference between the two ( called the residual ) in green. The residual has been magnified some 100 times so is actually rather smaller than shown. The properties ( magnitude and phase ) in question are of one main feed-forward circuit for the Hydraulic External Pre Isolator - the big cradles that hold crucial IFO components ( which is pretty much all of it ) off the floor and reduce transmission of vibrations to within. If you recall earlier discussion about control circuits then the magnitude is how much and the phase is when, both with respect to some input amount. Roughly speaking feed-back is a correction for something that has already happened, feed-forward is prediction of something that is due to happen.

Quote:
ASIDE : Suppose you are running past and I throw a ball in the direction of where I think you will be when the ball arrives, that's feed-forward. Once I've thrown it I can't correct, so success depends on accurate knowledge of how you will move and what happens to balls in flight. If you change your motion or a wind gust comes along, then the ball will miss you. ( Bullets are fired this way - ballistically ). If I could, mid flight after the throw, correct the motion of the ball depending upon whatever factors cause balls to miss moving targets, then I have feedback. You zig, the ball zigs. You zag, the ball zags. ( Air to air missiles are fired this way - guided ). You can use both feed-forward and feedback together, think of it like an initial guess followed by amendments depending upon how things are going.


So this is a predictive model to account for ( near ) future movements. The reason for showing you this is :

Quote:
As you can see, with the 20th order IIR, we can match the FIR response to within 1% at all frequencies.


That's the twentieth order! So if x is some quantity of interest then x^[0] = 1 ( zero-th order ), x^[1] = x ( first order ), x^[2] = x squared ( second order ), x^[3] = x cubed ( third order ) and so on. Hence x^[20] is twentieth order. So to get the suppression of vibration ( from outside to within ) in a feed forward system that predicts future movements to less than 1% error, you need a model that includes 20 powers of the input(s)! Related figures are quoted to 15 significant digits after the decimal point. This means that they understand the behaviour of the components very, very well. WOW, this is hard core .... :-)

More on degrees of freedom ( ASC = Aligment Sensing and Control ):

Quote:

ASC dof reconstruction
I wrote a little script (attached) to double-check that the ASC output matrix
is doing what it should - ie. when we think we're exciting a particular degree
of freedom, the script verifies that the mirrors really are moving in such a
way to produce that degree of freedom. It turns out that they mostly are -
other dofs are excited by as much as 30% though. The accuracy of this measure,
however, depends on the accuracy of the optical lever calibration.

The idea is that theta_dof = C^-1 * theta_mirror, where C is the control
matrix. Using recorded optical lever signals during a sensing matrix
measurement, you can calculate the exact angular motion of each mirror, the
theta_mirrors. Measured at 9.7 Hz and in units of urad, the reconstructed
theta_dofs for pitch are:
[pre] DU CU DS CS RM <-- dof excitation
1.0e-05 *

0.7183 -0.0562 -0.2464 0.0742 -0.0077 theta_DU
-0.0333 0.6832 0.0832 -0.2312 -0.0050 theta_CU
-0.2385 0.0457 0.7936 -0.0431 0.0064 theta_DS
0.0728 -0.2410 -0.0488 0.7738 -0.0018 theta_CS
-0.0066 0.0771 0.0019 -0.0302 0.6393 theta_RM[/pre]
You see, for example, that a DU excitation at 9.7 Hz produces 7.2e-12 rad of DU motion AND 2.4e-12 rad of DS motion. It is nice to see that an RM excitation only moves the RM. Yaw results are similar. Changing the 0.87 output matrix values to 0.91 did not have a substantial effect on these results.


The matrix as shown relates inputs ( 'excitation' ) to outputs ( 'theta' ) for the distinct degrees of freedom ( 'dof' ). I've highlighted some entries, the diagonal ones to show the desired result - nudging one degree only affects a single mirror angle - but as you can see from the orange non-diagonal entries other degrees are affected significantly too. As the author indicates, how well this works to keep the IFO aligned depends on how well known is the performance of the actuators.

Here's a spot of plumbing :

an upgrade of the chillers lines that keep certain lasers cool - the ones that help adjust the shape of the mirrors by selectively heating different areas [ Thermal Compenstaion System ]. About 4 gallons per minute at around 40psi passes through to keep the laser temperature stable. Otherwise fluctuations would effectively inject an unwanted signal into the interferometer. From the point of view of a photon in the lasing cavity the energy of an approaching atom, in an excited state, depends upon the ( gas ) atom's velocity. Or put another way the frequency of the photon from the atom's point of view depends upon dopplering. In any case a hotter laser has higher atomic speeds, and a greater spread of speeds, than a cooler one. This affects the odds of spontaneous emission of a photon from an electron in an excited atom transitioning to a lower state. So the laser ( peak ) power varies with temperature, in a complex way including other factors. From this document I've discovered this plot, an example showing how variable laser power can be with temperature and time :

It's perhaps easier to keep the temperature stable. Other readings of mine seem to indicate a desired range of less than a tenth of a degree Celsius/Kelvin. So in effect the laser is in a 'cold bath' of continuously replenished fluid to carry away any thermal energy generated from it's operation. The chillers ( refrigerators ) themselves are some distance away, in isolated enclosures, but that still leaves noise travelling up the pipes. So there are design elements that improve laminar ( non turbulent ) flow.

Here's some variant opinion regarding coax/BNC :

Quote:

Does this confirm my old training about coax connections?
I am impressed (and feel a little vindicated). I was always taught to terminate RG-58 connections to 50 Ohms, given its characteristic impedance as a transmission line. I have been amazed at the somewhat cavalier way things are terminated and tee-ed for timing signals in LIGO cabling. Clearly, we can do better.

I also twitch a bit when I see differential signals sent on coax. Don't they make good differential connectors (instead of having to remember to carefully insulate BNC connectors)?


Quote:

not in this case
In this case, we've just terminated the input to the ADC with 50 Ohms. We're not terminating the cable or matching impedances in this case.

In general, its not a good idea to put 50 Ohm terminators all over the place since most of our circuits don't have output buffers that can drive 50 Ohms. However, all of the RF circuits should have this and maybe all of the timing connections should use line drivers / receivers.


And where the snagging of a ribbon cable in a rack case door cut three channels :

Here's a great screenshot exhibiting the complexity of the control circuits I've been alluding to :

which gives the operators realtime knowledge of the moment to moment adjustment of IFO aligment.

More later. Bye for now! :-)

Cheers, Mike.

( edit ) A third type of control scenario is 'open loop' which is essentially just like an amplifier, but with no prediction or correction involved. Power assistance on your car's brakes ( not ABS ) or power steering are like this - simply magnifying a human effort.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

ML1
ML1
Joined: 20 Feb 05
Posts: 334
Credit: 55,037,852
RAC: 3,644

Oooer, hope all is well

Oooer, hope all is well DownUnda!

Pause, offline, or other?

Regards,
Martin

Powered by: Mageia5
See & try out your OS Freedom! Linux Voice
The Future is what We all make IT [url=http://www.gnu.org/copyleft/gpl.html](GPLv3

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

RE: Oooer, hope all is well

Message 96600 in response to message 96599

Quote:

Oooer, hope all is well DownUnda!

Pause, offline, or other?


It's OK. First anniversary of our firestorm, so a busy time for a local doc ....

Thanks for asking! :-)

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

Time get back into the

Time get back into the saddle. I won't try to catch up what's been missed. Hanford :

Nice bit of humour ( after a shift with a mere 8.3% duty cycle ):

Quote:
High winds continue to prevent locking. The LHC has managed collisions at 7 TeV, however.


and some trouble with the Control and Data Acquisition System ( CDS ), with curious behaviour of the underlying RAID array:

Quote:

hanford2 raid
Powered down hanford2, performed a system diagnostics
on the raid disks, they all passed (wait for it...).
Also split the two raid PS between UPS and Fac pwr.

Powered hanford2 back on, and its software raid marked
the sdi disk as failed. Moved the failed disk to hanford3's
raid and verified with its sys diagnostic that the disk
had mysteriously failed soon after it was checked.
So hot swapped a spare in its place, and found the raid
rebuild had restarted because of this, and the new spare
is now also marked as bad!

So I'm keeping my hands off hanford2's raid until it has rebuilt
(in four days time) and will then revisit the "bad" spare.
The CDS file system may be slower as the raid is being
rebuilt on-the-fly.

If you go looking for problems they will appear!


Now to boggle you all, check out an excerpt from this AdLIGO design document

Quote:
Experience to date with LIGO I has shown that any data that are acquired are required to be archived indefinitely. We will use this same data model as a conservative estimate for Advanced LIGO requirements. In this model, all data are acquired and stored for several weeks on-line in a disk cache at the observatories that is shared with the CDS LAN to permit real-time data access from the control rooms. The data are also ingested into the RAID cluster data array capable of storing ~ 2.5 PB on the cluster disk array. This is sufficient to accommodate more than 1 year of on-site data at each observatory (for all interferometers). Data will be streamed over the WAN to the main archive at Caltech, where multiple copies will be made for backup. Reduced Data Sets (RDSs) in this tapeless model can be produced wherever it is convenient (for initial LIGO the full raw data are initially only accessible at the sites, where all RDSs are created). The experience in initial LIGO is that several stages of RDSs are desirable, each reducing the volume of data via channel selection and data downsampling by a factor ~10X. As shown in the table, accounting for a 300% backup of archived frame data, Advanced LIGO will require a ~ 1.8 PB/yr archive capacity.


With Petabyte/Terabyte/Gigabyte/Megabyte/Kilobyte/byte each being 3 orders of magnitude apart. Copies fair enough, but why 3 in particular? An old sailing adage : "when you go to sea, take one clock or three...". Well if you had only two copies and for whatever reason they differ, and you want to correct that, which do you choose as the 'right' one? A far easier decision with three, if two of three agree, reconcile the third to the other two. Of course two could be 'wrong' with the lone third being the 'right' one, but that is far less likely ( using reasonable assumptions ) than the converse. By extension then, are odd numbers of copies are superior to adjacent evens? So particularly, is six copies versus five an equal improvement compared to five copies versus four? :-)

Now with high winds ( peaking at 60+ mph ) you get unusual problems :

Quote:
2 Praxair trucks, including one getting stuck in tumbleweed debris on X.


Cheers, Mike.

( edit ) 'Reduced Data Sets' refers to the fact that not all data segments are equally 'valuable'. We know that the moment to moment operation of the IFO's, with all the various sources of noise, causes different sensitivities of the devices overall to any passing gravitational wave. Data segments are still collected but are flagged if there are IFO problems, so that later one can weed those out if one chooses.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Ver Greeneyes
Ver Greeneyes
Joined: 26 Mar 09
Posts: 140
Credit: 9,562,235
RAC: 0

With the mention of the LHC,

With the mention of the LHC, then the talk about data acquisition and storage I got curious: to what extent, if at all, can they make use of the GRID, and is it expected that this use will be expanded in the future?

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

RE: By extension then, are

Message 96603 in response to message 96601

Quote:
By extension then, are odd numbers of copies are superior to adjacent evens? So particularly, is six copies versus five an equal improvement compared to five copies versus four? :-)


I'd assert that six copies versus five doesn't offer anything over five versus four. You'd have to go to seven copies for any extra boost in reliability.

My logic : if I have three that agree with the remainder not, then whether that remainder is two copies or three is pretty equivalent. Actually the case of 3/3 split for 6 in total remains unresolved, whereas 3/2 in a total of five would be. The cases of 4/2 ( total of 6 ) and 4/1 ( total of 5 ) aren't a challenge.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

Now it looks like Livingston

Now it looks like Livingston has had very good duty cycles and decent inspiral range. They've also had some trouble with data collection, and nicely managed by redundancy :

Quote:

Frame-writing disabled on FB1 to clean up overfilled frame disk
Igor reported that about 23:00 yesterday they were seeing corrupt frames from at least one frame-builder. The FB0 frame-builder frames looked fine, however the disk with the FB1 frames was at 100%. I was unable to access the disk normally with it in that state. I have thus disable frame-writing and reading on FB1 by editting the daqdrc files for fb1w, nds1, nds3 and re-starting the FB1 DAQ. This is still a valid running state, so after this change, the scimons can attempt going back to science mode.

I will meanwhile work to clear the disk. I have so far unmounted it from FB1 so I can work on it when not shared


Now I've been reading somewhat avidly on the topic of General Relativity in the hope of understanding what's what. I'll share a minor epiphany with you, if you'll suffer it! :-)

I've been pondering on what is actually meant, in practical or theoretical terms, by spacetime "curvature"? What entity is it that is "bending" or "warping"? For that matter, what is meant by "straight"? Here's my run at it :

Start by throwing out, if you can, any ideas/preconceptions about either space and/or time. Set your mind to a blank canvas of nothing and then see how we can 'construct' a universe that is consistent with the principles of GR. You see we tend to be, naturally, fixated upon the everyday world. Thus we need to adjust to the notion that our usual experiences are in but one corner of the universe, and that 'foreign' notions may reign elsewhere but which none-the-less approximate quite well here. For this discussion ignore non-gravitational forces, lets see what gravity does alone.

Think of a particular point in space, at a particular time, but by itself and not described with regard to any greater framework of description ( co-ordinate system ). For definiteness say this point is when & where a particle collides with another, a so-called 'event'. It is thus a 'real' thing, not abstract. How do I get to a nearby point? ( there must be a nearby point, else the universe has no extension ). Refer locally to a set of vectors that exist for this point alone ( technically called the "tangent space" at the point ). These are pointers to nearby points. A bit like road signs at the main intersection of a village - there aren't lines along the road to follow, but a set of choices to make about 'where to next?' for reaching another close-by village. Trot along the chosen road to the next village. What now? Well there is another set of vectors again ( another tangent space ) to guide any further travel. Et cetera ..... the image in my mind is of the French bocage country where you can't see over hedges and whatnot to gain a 'distant bearing'.

We are used to vectors going between known points you see, according to some co-ordinate system. So if I have a point A and a point B, I draw 'a line' from A to B and stick an arrow or some such marker to specify which end of the line ( segment ) is the 'beginning' and which is the 'end'. This is the standard vector idea, which implies all sorts of things - like being able to pick the vector up and drag it somewhere else ( retaining it's qualities like length and orientation ) and having it represent the general idea of displacement upon a known/predetermined background.

Not so with these local ( point dependent sets of tangent ) vectors, which aren't displacements, aren't draggable, don't really have a length in the commonly meant sense, but effectively are road signs to navigate through space and time. For the mathematically inclined they are derivatives implying the notion of 'most direct route'. However there isn't a standard set of 'direct routes' that work for all places, in the sense that the use of a compass would offer. You have to use one and one set only for each and every point in spacetime.

Quote:
[aside]
To be pure I must say these nearby points must be infinitesimally separated - which is a way of saying that for any given pair of 'nearby villages' I can always insert another in between! Each time I do I must have another set of tangent vectors - signposts at the village intersection - to serve that new village alone. That implies an infinite set of villages, but if that worries you then just ignore this aside! :-)
[/aside]


This no doubt seems are rather obtuse way of going around things. Why bother with such a baroque approach?

Firstly this does work for our 'everyday' life. It implies all the things we enjoy about our intuitive Euclidean understanding of the world. I replace all traditional 'point to point' vectors with a trail of tiny little steps, and I iterate the same technique at each step. I'm at a point, I look up the tangent vectors for this point, select one and go that way to a nearby point .... at this new point I examine the tangent vector set, etc. Are we there yet? :-)

Secondly, I don't need to make any global statements. I'm only ever referring to local decisions.

Thirdly, and this is the true majesty of this approach, I can describe any 'reasonable' universe. The 'curvature' concept becomes a statement that I can have more than one 'direct route' between well separated events. So a transit between a 'Berlin' event and a 'Paris' event ( just names for this analogy ) can have several routes each of which are 'minimal'. I can set off in different directions from Berlin, taking a village-to-village navigation technique with each initial choice, and still wind up in Paris with no reason to 'prefer' one total route over another. In a classical sense I would say "parallel lines have met somewhere in the distance". Is this not curvature or warping or bending?

Quote:
[aside]
You might consider gravitational lensing as a good example here, where an intermediate and unseen galaxy brings otherwise divergent light paths back to intersect one another. But be careful as it is spacetime we are describing - it has been noted that some arcs of light in a gravitational lens image vary similarly in time, but with delay!
[/aside]


Fourthly, I can make global statements if I want. I examine various paths which differ slightly from one another - two 'close' paths near a black hole horizon, say - and see the result of such variation. This is a way of saying that I can construct a co-ordinate system, rather than assume one, by looking at the totality of spacetime paths. Reference frames then emerge as a result of applying local rules ( the tangent vectors especially ), not local rules deriving from a larger framework. The 'metric' of spacetime is the characterising of the variation of tangent vector sets as you move from point to point between events.

Just ignore this post if it doesn't make sense ..... :-) :-)

Cheers, Mike.

( edit ) As applied to a LIGO detector the two key events are :

(A) two identical photons separating at the beam splitter in the corner station, and

(B) the same two photons returning to the beam splitter having each experienced paths in a different interferometer arm

with the phase difference ( as evidenced by the probability of photo detector capture ) mapping the relative topology of the arm paths.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

tullio
tullio
Joined: 22 Jan 05
Posts: 1,994
Credit: 31,846,322
RAC: 7,759

A good introduction to

A good introduction to General Relativity by intuitive reasoning is contained in "Turtle geometry" by Harold Anderson and Andrea Disessa. The LOGO turtle is a local coordinate system in a spacetime without a general coordinate system. It is a very enjoyable book if you like LOGO.
Tullio

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

I've been having a look the

I've been having a look the 'glitch' business. The logs show a shift by shift glitch report, like this :

Quote:

Glitch report - Owl
1. 954687146.4414 (Peak: 954687146.4414, Start: 954687144.5096, End: 954687148.4953, f: 777.90, SNR: 201.69)

DQ flags:
H1 DMT-SEVERE_OM1_OVERFLOW 1 In-loop OM1 overflow readout
H1 DMT-OM1_QPD_OVERFLOW 1 OM1 quad photodiode overflow
H1 DMT-OM1_OVERFLOW 1 OMC 1 readout/conntrol overflows
H1 DMT-LIGHTDIP_6_PERCENT 1 Dip in interferometer arm power by 6 percent
H1 DMT-BRMS_SEISMIC_Y_30_100_MHZ_LOWTHRESH 1 Elevated (low threshold) seismic noise in the Y direction at the end-Y station in the 30-100 mHz band
H1 DMT-INSPIRAL_RANGE_STDEV_GT_1_MPC 1 Standard deviation of inspiral range exceeds 1.000000 Mpc over last 10 minutes

This is the beginning of the Papua New Guinea earthquake right before OWL shift ended

2. 954686162.0000 (Peak: 954686162.0000, Start: 954686161.7448, End: 954686163.0667, f: 79.55, SNR: 106.51)

DQ flags:
H1 DMT-LIGHTDIP_6_PERCENT 1 Dip in interferometer arm power by 6 percent
H1 DMT-INSPIRAL_RANGE_STDEV_GT_0P50_MPC 1 Standard deviation of inspiral range exceeds 0.500000 Mpc over last 10 minutes

The second of a series of 3 glitches that light up many channels. Some LSC channels (namely LSC-POB_I and Q and LSC-MICH_CTRL) have almost exactly the same glitches as DARM. The OMC channels all (and many ASC channels) have a series of periodic glitches that match the timing of the 3 in DARM.

3. 954675507.1914 (Peak: 954675507.1914, Start: 954675507.0503, End: 954675507.3539, f: 1034.80, SNR: 100.95)

no interesting DQ flags. Typical short, hard, uninteresting glitch


With a glitch being defined ( see here ) as :

Quote:
any short-duration noise transient in the gravitational wave channel as well as transients in auxiliary channels.


The glitch reporting and study is overseen by the glitch group, a subgroup of the Detector Characterization Committee, with members plucked from various other groups that design, commission and operate the detectors or those that deal with the output. Glitches matter because some of the astronomical signals happen over a similar time span and we don't want any terrestrial causes mimicking that. Obviously if glitches can be reduced - because their origin is understood and correctable - then the IFO's may work better. But even if not that, then those data segments produced at times when glitches occur can be identified. Which helps in selecting better quality data for any analysis. There would be more than a few non-correctable causes, though. Power supply problems, earthquakes, someone stubbing their toe .... :-)

Quote:
[aside]
This sounds reminiscent of trying to find and remove the squeaks from a car dashboard ! Never, ever, buy an Aussie GM/Holden Commodore. Or if you do then simply set it on fire as soon as you leave the car dealer. It's just quicker that way ..... :-)
[/aside]


My feeling is that glitches aren't anywhere near as important for the continuous wave sources that we are trying to detect, compared to other efforts. We integrate over rather longer periods than a few seconds so glitches would be somewhat less likely to fool our searches - particularly if they have no real periodicity.

I wonder if there may be some analogy with chlorophyll here? No single photon can/will trigger a chemical reaction, however the electrons run up and down a band within the molecule absorbing successive photon energies. The reactive focus is where that is released. Think of a kettle on a slow heat, it will eventually boil on account of accumulated input. Or the boiler with a safety valve on a steam train. So ( recalling an earlier discussion of resonance ) an interferometer may simply have mechanisms of sudden energy transport, at certain thresholds, exhibiting as glitches. Resonant bar detectors are known to integrate effects in that way. Which is not only bloody annoying, but would make the glitch events slaved to history ..... that is, the response to a signal depends very sensitively on the state of the detector prior to it's arrival.

Now while the surface of the Earth is evidently an especially noisy place, LISA could still suffer rather too. Far more high energy particle interactions for starters, not having the benefit of any of Earth's magnetosphere shielding it.

As for GR : while we may well have this 'nearby village signposts' business, and the Einstein equation ( well a group of 16 actually, but only 10 are independent ) to specify how they ought behave one-to-another across our spacetime landscape, we are barely into the problem at all. GR can describe entire universes other than this one, and not simply different parts of this one. So that becomes more than a matter of expertise at solving or approximating solutions, as clever as one can be, but a more wider concern of initial conditions. What you start with certainly determines what happens later on. GR can't tell you what those initial conditions are, only how stuff evolves. In any case GR is classical so really has nought to say about extremely high density/energy/quantum scenarios - except to label them as singularities. Like the Big Bang or the center of a black hole. Ironically Einstein loathed the quantum concept, and notably probabilistic interpretations, that he helped deliver to us. So no surprise GR has no quantum component, not-withstanding having done spectacularly well at those non-quantum scales it has been tested upon.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,043
Credit: 105,828,719
RAC: 50,458

Looking at Livingston I see

Looking at Livingston I see an entry containing :

Quote:
Sigg-Sidles instability at 10W
I directly measured this evening the Sigg-Sidles instability in the arm cavities. The first page of the attached plot jumps straight to the punchline that measurement meets theory .....


Eh? Upon some research I discover this great article that tells of a prediction ( by Daniel Sigg and John Sidles ) that higher laser beam power in the interferometer arms can cause instability. Since AdLIGO is really going to ramp up that power then this becomes an important concern. I'll try to paraphrase the problem/concept, by my usual recourse to analogy ..... :-)

- imagine one is standing 'squarely' in a neutral relaxed fashion. So each foot will be more or less below the hip on each side, arms are hanging loosely with palms facing the upper/outer thighs. My shoulders will be more or less above those vertically. Because of the way humans are built this is a 'settled' posture, meaning it takes some minimum of energy to maintain it ( given that one is standing upright at all ).

- look down from above and consider rotation about the core of the body. That is around a vertical line that goes from midway between the feet to the middle of the top of the skull. So if my right shoulder moves forward, my left shoulder will move back and the flat area around the centre of the front of my chest will face over to the left somewhere. Conversely if my left shoulder moves forward, my right shoulder will move back and the flat area around the centre of the front of my chest will face over to the right somewhere.

- if I have either of those rotational positions the various tissues in my body will resist and tend to pull me back ( untwist me ) to the neutral square-on stature. Try it yourself, you will probably feel some tension especially in areas between the bottom of rib cage and the top of the hips either side. So there is a natural tendency to restore any deviation from neutral position. This is typical of many dynamical systems in physics, like a pendulum for instance.

- but such systems are rarely undisturbed, indeed it would be pretty boring physics if not. In this case let's consider two people facing each other, within reach, in their respective neutral positions. Call them Bob and Alice. Alice pushes ( using either her right or left hand ) on Bob's right shoulder. This has two effects ( remembering Newton's action and reaction ) : Bob's right shoulder will move back and so too will Alice's shoulder ( left or right depending on her choice of pushing hand ). If it was just one push then both Alice and Bob's bodies will return to the neutral positions, maybe with a little bit of oscillation back & through the neutral positions if the shove was hard enough.

- to makes matters interesting let's throw in some 'tit for tat'! A push provokes a push in reply, and that deserves a push, and yet another .... what happens now in terms of the movements of Bob and Alice has a number of variations/patterns depending on the detail of the manner of response.

Come back to LIGO now, but remember the above analogy. Bob and Alice are mirrors at opposite ends of an arm cavity. The wire that suspends them is the body tissue that restores toward the neutral position. The centre of each mirror is the middle of the chest, with the outer mirror rims being shoulders. So what constitutes the pushes from the hands? The photons that travel between the mirrors :

don't worry too much about all the symbols, just focus on the bold double arrowed line marked P. That's the photon beam. Also note with these concave mirrors that the light will tend to be reflected back to the side that the mirror is generally facing. Bear in mind that many round trips of the photons will occur during the time that the mirror alignments change only a little bit. Thus the mirrors are moving in 'slow motion' compared to those zippy guys. Here is one type of tit-for-tat that can be played :

which is fairly harmless in the sense that the photon beam tends to push both mirrors back into alignment. However this one :

doesn't. It makes the angular deviations greater. The photon beam pressure works against the forces restoring the neutral position. You can imagine how this worsens as we flood more photons into the resonant cavity between the mirrors - say by making the recycling cavity better, or increasing the laser power at source, or allowing fewer photons to escape at the dark port. This behaviour must be suppressed, else the interferometer alignment will be lost.

w0 is the frequency that the mirrors will generally oscillate at without any photons around - so that depends on the stiffness of the restoring forces from the wires that suspend it, the mass ( and mass distribution ) of the mirrors etc.

w+ is the frequency of oscillation if the photon hits assist re-alignment. This is a higher frequency as angular deviations are restored quicker ( shorter period ) when the light is helping.

w- is the frequency of oscillation if the photon hits resist re-alignment. This is a lower frequency as angular deviations are restored slower ( longer period ) when the light is opposing.

Managing this aspect of interferometer control is achieved by placing an extra component to dampen :

which has been measured to work very well indeed :

I think this is a pretty significant piece of work. Now and for the future.

Cheers, Mike.

( edit ) Recall that the Wave Front Sensing allows us to detect angular movement of the beam off centre, so you'd plonk the control of this Sigg-Sidles behaviour in with that control loop. Also here's the rough size of the effect :

Quote:
Light arrives at the recycling mirror with a power of 5 watts, but inboard of the recycling mirror the effective power level hitting the interferometer's beamsplitter is 250 watts. Inside the resonant cavities that make up each of the long Michelson arms, the power level is 12 kilowatts. Thus, the radiation pressure on each of the arm mirrors is 10^(-4) newtons. Incident on a 10 kg mirror held in a pendulum suspension with resonant frequency of 0.5 Hz, this force would cause a displacement of 10^(-6) m.


the magnitude of which has been handled by existing control elements, but with the coming of AdLIGO extra abilities are needed.

( edit ) w-, w0 and w+ are way low for the range of gravitational wave frequencies that LIGO has best sensitivity for. So the main issue here isn't injection of signals into that range ( though I guess some harmonics might reach that ) but whether the interferometer remains in lock at all.

( edit ) Those mirror diagrams above aren't the whole modes, just instantaneous snapshots of typical positions during them.

So for the 'stiff' mode you could wait a little while and see an arrangement in a mirror reflection about a horizontal line. The key point is that the members of respective shoulder pairs ( Alice's left shoulder and Bob's right PLUS Alice's right shoulder and Bob's left ) have a more or less constant separation throughout the oscillation cycles. The photon beam axis ( bold double arrowed line marked P ) will essentially rotate around some mid-point between the mirrors. Like a see-saw.

Whereas for the 'soft' mode, by similiar symmetry, has the members of the above shoulder pairs moving toward and way from each during a cycle of oscillation. The photon beam axis will walk up and down the mirror surfaces but stay more or less parallel to the central line between the two. Like the floor of a lift ascending and descending.

( edit ) I should add that this original paper yields the idea that the presence of the photon beam effectively alters the 'stiffness' of the mirror suspensions, hence the terms 'stiff' and 'soft' to describe the modes. Bear in mind that at any given moment the Sigg-Sidles mechanism may be only one of many influences upon the mirror movements.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.