Detector Watch S6 V2

Dan G.
Dan G.
Joined: 10 Feb 07
Posts: 39
Credit: 97,382
RAC: 0

Thanks a lot for all the

Message 95185 in response to message 95184

Thanks a lot for all the great references. I've also been trying to get up to speed with general relativity in order to better understand GW from a theoretical standpoint. I'm using this Book. I wonder if you've heard of it?

Also, I'm also a science journalist and I'm wondering if I could interview you about your interest in physics. If you go to my
Blog and use the e-mail address there to contact me, I can forward my questions.

Thanks, and keep up the great work!

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,525
Credit: 281,531,356
RAC: 40,969

[ Sorry for the gap, been out

[ Sorry for the gap, been out and about a bit recently plus the DSL port at my local exchange augured in ... ]

Quote:
Also, I'm also a science journalist and I'm wondering if I could interview you about your interest in physics. If you go to my
Blog and use the e-mail address there to contact me, I can forward my questions.


It will be a pleasure Dan! I will indeed! :-)

I get the feeling that things just ain't going right. There's a lot of lock loss ( or failure to gain ) at Hanford - due to seismic activity - which isn't typical for the site compared with previous years. Check this out as a likely cause ( yep, the little coloured blobs are people ) :

with this terrific explanation/commentary from one of the scientists ( Robert Schofield ) :

Quote:
I took this picture early last weekend at the Oregon coast during the high microseism. This water fell just beyond the people, so they were not hurt, but a few waves later, an ambulance had to carry away the incautious. The huge microseism at LHO had a period of 7.47 s at the time of this photo. At the same time, I counted the expected 15 seconds between waves: the microseismic peak is at half the wave period because the microseism is not produced in the very thin zone where waves crash into the shore, but by standing wave fields (at half the wave period) with sizes comparable to the ~ 50,000 m wavelength of the microseismic peak signal in the ground. The center of mass of the ocean below a standing wave field moves up and down (at half the period of the waves), coherently applying force to the ground over a ground-wavelength scale region, thus improving coupling. The standing wave fields are likely the combination of the incoming and reflected waves.


That is, the continent is getting regularly thumped by a big fist and it resonates. What a drum kit! :-)

Also various components seem to be either failing or going off specification, in no particular pattern/area that I can discern. The reflective memory has had another hissy fit. Now :

Quote:
Segment 690 with Plane Bump
At about 02:43:43 utc, an airplane flew right over us, raising the plane peak on DARM_ERR spectrum (about 80 Hz and moving), checked planemon, and yes it is there.


So here's a quiz question for you all. What rocks n' rolls at 80Hz on an airplane in flight? ( The log doesn't indicate what sort of plane ). Answers/suggestions on one foolscap side please ...... :-)

Now Livingston ( my fault, I mentioned a quiet hurricane season earlier ) is flapping in and out ( more out ) of science mode as the Gulf is revving up. Hurricane Ida :

so apart from watching how the IFO responds to this, not much science was done. But the operators haven't been idle. Far from it! They've been poking around and doing a whole range of stuff while the IFO has no hope of locking. One has turned to poetry ( thank you Mr Tom Evans ):

Quote:
When Ida blew across the Gulf so deep,
And operators tried in vain to lock,
The micro-seismic noise did upward creep,
And arms both X and Y did rock.
The sci mons could but check their data plots,
When wind blew many weaker branches down,
And weather radar showed us naught but blots,
While rain soaked this Louisiana town.
But now the seismic noise has passed its peak,
And the alignment scripts have done their task,
It looks like Ida has become quite weak,
And so, with hope, we can at last now ask,
"Before the light of this fine day has passed,
will we be able to acquire at last?"


Sorry Dan, I don't know of that book. Let us know how it goes. On the up side, I've done my yearly pre-Xmas major book buy. I've got Meisner/Thorne/Wheeler's Gravitation and Sean Carroll's Spacetime and Geometry: An Introduction to General Relativity. I've read about the first dozen pages of each and I am stoked. Did you know that another way of saying General Relativity is geometrodynamics?

Cheers, Mike.

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

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,525
Credit: 281,531,356
RAC: 40,969

Well, um, it looks like the

Well, um, it looks like the OMC problem at Hanford may be software related ( my bold highlights ) :

Quote:
"... OMC has an UP script but not a complete DOWN script!!! ... This completely explains why Gerardo's trick of running the UP scripts with an unlocked IFO sometimes fixes the IFO! "


Quote:
OMC wouldn't lock to the carrier (or sideband) via PZT/Heater changes, and seemed to be driving it off rather than on. Recalling the past instances of OMC LSC Phase jump, most recently on Tuesday, flipped the sign of the phase and it locked right up. Used ezLSCPhase to remeasure and it set the phase to -131.3 degrees (starting at -129.5, ran again to confirm). Used to be 129.5, so it's jumped by a lot. This is the second time in 48 hours, which is also suspicious. The only mention of recent OMC work is in MichaelR's shift summary about Richard testing timing cables, but the boot log also notes reboots of om1 and om2 at (local) 10AM and 3PM, which are not commented on in the day's posts that I can see. This matches Keita and Nic's suspicion that front end reboots can change the phase.


.... which if I'm reading correctly ( ahem/cough ) is a good thing because when fully understood will probably be a simple fix. I'm assuming 'front end' means software control of the relevant degrees of freedom. Phase flip basically means it zigged when it should have zagged, like a left hand turn instead of a right when driving. You wind totally up the wrong street. The Output Mode Cleaner, by using resonance in a cavity, is tuned to preserve the 'carrier' ( base laser frequency, not the sidebands ).

My guess is that this could also be related to the 'fringe wrapping problem'. The IFO arm lengths don't have to be identical for lock to occur. They just need to be some odd multiple of half phase different. Compare a photon up one arm with regard to the other. The configuration where the X arm is say 1000 cycles round trip but the Y arm is 1000 + 1/2 then you'll get nil at the dark port. But X arm at 1000 cycles per round trip and Y arm is 1001 + 1/2 per round trip, gives the same effect. And by extension Y arm at 1002 + 1/2, 1003 + 1/2 ... or the other way, 987 + 1/2 for that matter ( X arm staying at 1000 ).

[ NB - the real number of cycles per arm trip is a humongous figure ]

So the phrase 'fringe wrap' is where the difference has moved from say 1/2 cycle to 1 + 1/2 cycles. Fringe is sort of an historical term from the days when the light/dark bands on an interference pattern had to be human visible ( no photodiodes etc ). A 'fringe' was a full cycle from dark to light to the next dark, say. The 'wrap' is moving across the pattern either way. A fringe wrap occurs when the IFO was in lock at say 10 + 1/2 cycles differential b/w the arms and went to ( say ) 11 + 1/2 cycles. Again the real numbers are quite big, and the neat thing is we don't have to know the absolute number of cycles - or their difference - provided that 1/2 is there.

I think it was John Wyndham's Day of The Triffids where the main character temporarily loses his sight, losing track of time. While he knows what the clock says now and before, he doesn't know how many days have elapsed. The clock winds around past 12 of course. Phase is like that, you have an accumulated time but the clock face is cyclic.

Incidentally, my reading of Fabry Perot interferometers indicates one of their uses being to 'count fringes' in some circumstance. If the frequency/wavelength of light used is accurately known then this equates to a distance measurement - each fringe is one wavelength.

Even Newton was effectively able to do this with a gently curving, highly polished, glass lying upon a flat one. Think of the surface of a sphere touching a tangent plane. Shine a light on the two, look down upon it, and you get light/dark rings centred on the point of contact. The beauty of this is that you know the centre point is at zero separation ( or you hope so, with good polishing of the surfaces ). So you can count, like tree rings, outwards from the centre. So one dark ring to the next is a fringe wrap.

As regards aeroplanes flying over, some casual research of mine has revealed an approximately 80 Hz resonance in the fuselage, mode along the longitudinal axis of larger planes. Apparently this is studied because it is relevant to the behaviour of electromechanical couplings within it. So various sources of vibrational energy, basically the engines, will set it rocking. Most of the mass of a plane is usually the fuselage. That will generate a signal for the IFO to pick up provided it's close by.

Not a whole lot to say about Livingston - business as usual ( trains, planes, automobiles, waves, earthquakes, logging ... ), so external events bumping it very frequently. Any lock gained is relatively low power. There were a few longer continuous segments.

Cheers, Mike.

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

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

ML1
ML1
Joined: 20 Feb 05
Posts: 347
Credit: 86,292,162
RAC: 797

Good interesting details as

Message 95188 in response to message 95187

Good interesting details as ever!

Quote:
As regards aeroplanes flying over, some casual research of mine has revealed an approximately 80 Hz resonance in the fuselage, mode along the longitudinal axis of larger planes. Apparently this is studied because it is relevant to the behaviour of electromechanical couplings within it. So various sources of vibrational energy, basically the engines, will set it rocking. Most of the mass of a plane is usually the fuselage. That will generate a signal for the IFO to pick up provided it's close by. ...


OK, just to be clear on this one...

So large aircraft have their central bodies resonating at about 80Hz, excited by engine vibrations at whatever frequencies.

So is LIGO picking up the 80Hz due to the gravitational effects of a hundred tons or so of 80Hz wobbling nearby, or due to other modes such as acoustic pickup?

Regards,
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,525
Credit: 281,531,356
RAC: 40,969

RE: So is LIGO picking up

Message 95189 in response to message 95188

Quote:
So is LIGO picking up the 80Hz due to the gravitational effects of a hundred tons or so of 80Hz wobbling nearby, or due to other modes such as acoustic pickup?


Probably both. But far, far more so acoustic is my guess. Pretty impossible to tell without perfect isolation.

Cheers, Mike.

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

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,522
Credit: 669,780,951
RAC: 679,520

I wonder what would be the

I wonder what would be the number of combustions per second in a small plane prop engine. For example, with a 4 cylinder 4 stroke engine, there would be 2 bangs per rev and 80 Hz would translate to 2400 rpm, I think that could be about right ??

I haven't worked the numbers, but I have a feeling that if you could really see the (newtonian) gravitational gradient from a low-amplitude vibration of the hull of a big (==> high flying, say >> 10k ft ) plane you should also see effects from birds flapping their wings??
CU
Bikeman

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,525
Credit: 281,531,356
RAC: 40,969

RE: I wonder what would be

Message 95191 in response to message 95190

Quote:
I wonder what would be the number of combustions per second in a small plane prop engine. For example, with a 4 cylinder 4 stroke engine, there would be 2 bangs per rev and 80 Hz would translate to 2400 rpm, I think that could be about right ??


Quite right. A typical Lycoming ( very common brand ) is 180 HP ( 360 Cubic inch ) 4-cylinder 4 stroke ( 2 revs per bang per cylinder ), thus 1/2 rev of the crankshaft per bang for the whole engine. Cylinders horizontally opposed. 2400 RPM is indeed a typical throttle setting for level cruising, maybe 2500-2600 for climb. [ So 2400 * 2 / 60 = 80 Hz ] Nearly all types using it, say Pipers, have the prop directly on the shaft too - no reduction gearing. Generally two bladed props as well. Plenty of opportunities with this for driven resonances.

Quote:
I haven't worked the numbers, but I have a feeling that if you could really see the (newtonian) gravitational gradient from a low-amplitude vibration of the hull of a big (==> high flying, say >> 10k ft ) plane you should also see effects from birds flapping their wings??


The effect here is proportional to mass and inversely proportional to the cube of distance. Sooooo with ....

Mp = mass of plane
Mb = mass of bird
Rp = distance to plane
Rb = distance to bird

then

Mp * Rb^3 = Mb * Rp^3

( after alot of fudging ) try a 4 kg bird going past at 200ft would be equivalent to a 500000 kg plane ( heavily laden A380 ) going by at 10K feet?

5 * 10^5 * (200)^3 = 4 * 10^12 = 4 * (10000)^3

Yup, that works. :-)

Cheers, Mike.

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

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,525
Credit: 281,531,356
RAC: 40,969

Sorry for the outage, been up

Sorry for the outage, been up in Cairns checking out the local gravity field. Which is indeed pretty Newtonian, in case you were wondering. :-)

Hanford has gradually acquired more science lock as the week progressed. There is monitoring of the science data for 'glitches'. Here's one related to planes :

Quote:

1. 942895524.9883
(Peak: 942895524.9883, Start: 942895524.5589, End: 942895525.3537, f: 221.45, SNR: 70.29)
Nov 22 2009 03:25:09.9883 UTC

from the DMT DQ flags
H1 DMT-AIRCRAFT_VERY_LIKELY 1

a short (in time) and wide band (in frequency) glitch appears in the Gravitational Wave Data channels.


The quoted flag indicates some bit of software thinks it is an aircraft. It didn't upset matters for long : about a second ( End - Start = 0.7948 ). Those times are in GPS seconds ( zero is 00:00:00 on 6-Jan-1980 ).

As for fringe wrapping, perhaps one of the optics tables is the problem ( ISCT4 ). It 'floats' on an air suspension but can hit restraining stops. As a fringe is about a micrometer ( laser wavelength is 1064 nanometers ) it would be easy to flip across a few during a bump.

As for waves :

Quote:

Tomorrow Night, 11/23 in the evening, Seismic Will Be High, Locking Will Be Difficult
Using Robert Schofield's wavefront Seismic Predictor, I'm going to prophesize in the name of Aeolus, the god of the winds, that locking will not be possible tomorrow night.

If you run the model on the Eastern North Pacific, setting it to both "peak periods" and then "wind speeds," and then set the forecast to "animation," you'll notice that for tomorrow evening, a huge salvo of waves with low frequency resonant with uSeismic frequency and high winds will hit Washington.

Wave heights are low but I think the other factors are the ones that mattered .... peak period should be inversely proportional to wave energy coupling with seismic noise .... longer peak periods mean that the winds are driving the waves faster longitudinally into the shore.


Quote:
Mea Culpa: I Misunderstood
After much introspection and further study, I'm informed that ocean movements are merely indicative of slow progression of microseisms and the coupling with the ground is a lot slower than I thought. So if anything, microseism will simply rise slowly later in the week, which isn't as interesting.


meaning that more energy is imparted to the continental mass with certain wave behaviours, albeit with delay. This is typical of resonance - magnitude, frequency and phase dependence. However later there is a link mentioned to this article, the top photo of which clearly shows the bend in the Columbia River where Hanford is at. Thus the slow grinding of tectonic plates, as opposed to the quick slips during earthquakes, is a constant low level vibration. I visualise this as two pieces of sandpaper held together but slowly slid sideways. You'd get lots of little motions perpendicular to the plane of the paper as the particles of grit lift and fall over the ones stuck on the opposite paper - like a car tyre going over a speed hump

Livingston won't lock for very long and lowish power at that. But like Hanford improved as the week progressed - one long 7+ hour segment in particular. Now here's a real downer on the night shift :

Quote:
Skunk hanging in or around LVEA, smelled but not seen!?


Quote:
On a more random note as of just recently the control room smells like
there is a skunck around here.... Yup, a skunk!


I have the personal pleasure of not knowing what that smell is like. I'd assume it would be difficult in it's presence to focus on anything complicated - like typing, operating a gravitational wave interferometer, tying shoelaces and/or not vomiting. Can anyone from the Skunk Infested Climes confirm that for me? Is there an analogous scent to compare with, some mix maybe, or is it truly unique? ;-)

Cheers, Mike.

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

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,525
Credit: 281,531,356
RAC: 40,969

Now I thought I'd come back

Now I thought I'd come back to the topic of angular alignment. It can be a bit painful, so put your thinking caps on! :-)

It's all very well having little magnets that move the mirrors back and forth to adjust the gaps between mirror surfaces. That gives us resonance as one can fit the cavity to the phases required for light to superpose in the way we want. So at the power recycling cavity we want to keep the light in, at the dark port similiarly. Bung in some feedback loops and one can hopefully maintain some useful state of the interferometer, large disturbances not-with-standing. Here's a reminder of the degrees of freedom a mirror can have :

Most of these are constrained by the mountings, or at least have fixed offsets in the sense that they don't necessarily partake in the moment-to-moment adjustment of the interferometer. Bear in mind that while produced to very high tolerance the mirrors aren't necessarily as symmetric as the picture implies. So you'd think the roll mode wouldn't matter, but it might if the surface isn't perfectly machined. And nothing is ever perfect. Also the mirrors are not flat faced, they are shaped to effect a 'focus' at some distance from it's face. You'd expect the mirrors used in the 4km arms to have mildest curvature. With the 'type' of light that the laser produces, one gets an effect like this :

[ The reason for this is rather complex ( yup, quantum mechanics! ) but is related to the fact that the laser doesn't produce an exact frequency, and the aperture ( beam width ) is finite. ]

What if we can't hit the mirror at the other end of some given resonant cavity? At a few meters this isn't too hard, but at 4km that's a real challenge. What helps, naturally, is that we are using a laser, but there is some spreading of the beam. So we have the wavefront sensors ( as discussed earlier ) to tell us if we are on the mark, and the little actuators to adjust :

Now how do we deduce what is causing any misalignment, and thus what to do to correct it? We come back to the idea of 'degrees of freedom' again. Here 'degrees' doesn't mean an angular measure but simply some aspect we can alter. Think about just the yaw motion :

What is illustrated is the mirrors or combinations thereof which produce a particular effect on the output. You will see ( shaded light green ) a given mirror can be involved in more than just one mode of movement. The terms 'common' and 'differential' are used in an analogous sense to the length degrees of freedom discussed earlier. So 'Differential Input Test Mass Mode' for instance shows if one input mirror rotates clockwise, the other will be going anti-clockwise. 'Common End Test Mass Mode' shows that both end mirrors move in the same sense.

Now a real curly bit is 'angle to length error'. If you can imagine a mirror not quite aligned in the proper direction then the mirror up the other end is going to be hit off centre by the beam. A given photodiode ( in the wave front sensor ) doesn't 'know' this though, all it measures is an intensity. Because there is a slight extra distance to travel to reach the periphery of a mirror compared to the centre, then there will be a slight difference in phase between light hitting various points outwards from the central part. This will affect intensity - interference between photons within the beam now - and could thus mimic an intensity change produced by a length adjustment. So what is actually an angular alignment problem is sensed as a cavity length change. The interferometer's angular and length behaviours are thus coupled.

I haven't mentioned another effect - errors accumulate. So if one mis-hits one mirror which then reflects/transmits light to another/same mirror the problem compounds!

The way to manage all this complexity, is to have a 'global' view of the device because of the above concerns ( plus other stuff I don't really understand ). Thus the responses ( currents to the little magnets etc ) need to reflect that total state. Mathematically this means matrices. Don't panic! :-)

Matrices are a way of managing globs of information which have related sub-parts. So suppose I run a fancy taxi service, with some number of drivers and various limos. To match the drivers with limos for a days work ( times, places and customers ) I could resort to matrix use. I'd have a matrix ( rows and columns ) to represent the drivers, another for the limos, a customer matrix etc ... and I'd formulate relations b/w these ( matrix equations ) and crunch a suitable algorithm ( see linear algebra ). Here's a sample matrix :

The DETM/DITM/CETM .... are the angular modes in the previous diagram. The left side shows what port ( specifically what photodiodes ) is measuring what, and how changes is the modes affect the signal received by said photodiode. You'll note the biggest numbers ( 1731/1602 ) relate to the differential movement of the input and end test masses. In other words a small error in mirror alignment over the longest cavity - the 4km arms - for a zig instead of a zag, has the biggest effect on light received at the dark port signal. Notably the other 4 modes have no effect for that PD. In any case this illustrates the general idea that correction to any mirror must involve information from all photodiodes.

This above matrix ( think of it as a table of values like a spreadsheet if that helps ) is only one of many. Each and every degree of freedom is potentially involved in any photodiode current, and thus feeding back to possibly any actuator movement. That's what happens when you bounce light around a maze of mirrors in the hope of measuring unbelievably small distance changes - a fraction of a proton's width! :-) :-)

Cheers, Mike.

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

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Martin Ryba
Martin Ryba
Joined: 9 Apr 09
Posts: 48
Credit: 151,809,074
RAC: 36,993

RE: RE: 1.

Message 95194 in response to message 95192

Quote:
Quote:
1. 942895524.9883
(Peak: 942895524.9883, Start: 942895524.5589, End: 942895525.3537, f: 221.45, SNR: 70.29)
Nov 22 2009 03:25:09.9883 UTC

The quoted flag indicates some bit of software thinks it is an aircraft. It didn't upset matters for long : about a second ( End - Start = 0.7948 ). Those times are in GPS seconds ( zero is 00:00:00 on 6-Jan-1980 ).

Cheers, Mike.

Hmm, as a GPS geek, I find it entertaining they decided to use the GPS time scale as the log reference. Makes sense, as using GPS time avoids leap seconds chaos. Putting it in straight seconds as opposed to the standard week number/Time Of Week setting makes printing and math easy even though the numbers are big. I assume your logging time scale is coarse enough that the roughly 100 microseconds granularity is OK...you're at 13 significant digits already. When I'm crunching GPS signals to nanosecond levels, I need to keep the seconds field under 10^6, and use some care to prevent accumulated rounding errors.

"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)

Comment viewing options

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