0. Executive Summary: As a response to the lock losses observed here, we make an attempt to create WFS filters that allow us to convert the WFS system into strong man mode while locked. Here is only the statement that some filters were replaced (two days ago actually) and a description of the filters. Later we will post some measurement results for these changes.
1. Introduction: In order to prevent angle to length coupling from translating WFS noise into the LSC land, very aggressive stop band filters (shown in brown here) are used to cut the servo gains before the WFS noise starts to dominate DARM_ERR. The problem with these is that they add a lot of phase at lower frequencies and limit the gain which we can apply to the WFS.
Because of this limitation, during strong seismic events the angular degrees of freedom are not held strongly enough and are causing oscillations as shown in this entry. Here we will investigate the possibility of cutting the science mode for a little bit in order to change out the WFS filters for ones that we can easily ramp up the gain. In this manner we will substitute DARM_ERR goodness for beefy, burly WFS system that can stick out noisier seismic conditions.
The Wave Front Sensors are used to make sure that we are looking at the right parts of the light coming through the dark port of the interferometer. We want the face of the photodiode to be 'square on' with respect to any light coming through. Or equivalently we want any/all the photons we catch to be in the same phase. So if the photodiode is 'skewed' then we will be receiving photons that have a spread of phase. If the interferometer is otherwise properly locked and whatnot then we would get some photons that have returned precisely one half wavelength ( or odd multiple ) out of phase with respect to the other arm, some a bit less than half a wavelength, and some a bit more than half a wavelength. This effectively introduces an unnecessary length uncertainty, and behaves like a depth of field effect in camera optics. This is the 'angle to length coupling' mentioned above. We want to 'focus' as best as possible on an exact plane.
Michelson and Morley did this deliberately. They didn't have sensitive photodiodes alas! So they arranged matters so that they could view a spread of phases. It appeared as fringes, a series of light and dark bands. However what they were looking for was a shift in the fringe pattern, which they had some hope of detecting with The Mark I Eyeball Instrument, as the whole interferometer was rotated. Thus possibly demonstrating a directional dependence to light transmission, aether etc .....
Thus the WFS must participate in the adjustment of the interferometer, which is an active beast that continually self-adjusts via various control loops. One of these feedback mechanisms is the yaw filter on the WFS. Yaw is an angular deflection. In an aeroplane it is a left/right rotation about an axis through the centre of gravity and also perpendicular to both the wing axis and the fuselage axis. Or, say, if you look straight ahead and then rotate your neck to look left/right without nodding that's also yaw ( looking up/down is designated as pitch, tilting to left/right without nodding is roll ). Filter is the answer to the question - if the mirror is disturbed in yaw then what correction do I apply to the servos that hold the mirror aligned? The servos are little pushing/pulling devices to point the mirror.
There are two features to the correction : magnitude and phase. The magnitude is how much do the servos push. The phase refers to response to an oscillation, basically do you push in the same or opposite direction to the oscillation? ( this is not the photon phase discussed above ). Anyone who has had fun on a playground swing already implicitly knows all about this. It's like a pendulum. You can have someone push you of course, many do. But you can also 'pump up' the motion yourself by raising and lowering your center of mass with respect to the seat. Either way if the timing is right then the swing will increase, but you can time differently and suppress the motion too. As a kid I would spend heaps of time stuffing about with this aspect of swings - and quite naturally without any physics tuition! I bet you have too .... :-)
You can think of the filter as a black box with an input and an output. The input is how the mirror is deviating from some 'neutral' position, and the output is the magnitude/phase as above. You describe the behaviour of the filter by plotting what output happens for a given input :
Don't panic when you look at this. Basically for a given frequency of disturbance you look along the horizontal axis ( which is logarithmic to compress the display ) to find it and then go up to hit one of the colored curves. Then go from that point on the curve across to the left side axis and read off the value. The interesting area is the 30/50-ish Hz region.
The green curve is what the filter had been, which was giving trouble. Essentially when a jiggle would come along in this frequency range then the magnitude of the response would go down. The left scale for magnitude is marked in decibels. It's another logarithmic thing. Zero dB means no change in strength, positive ( up the axis ) is stronger, negative ( down the axis ) is weaker. So when things started rocking ( an earth tremor say ) the grip would relax, the oscillation would worsen and probably cause the grip to relax some more.
The blue curve is a new filter. It goes up in that frequency band and hence holds on harder. Plus if you note the phase curve ( where zero phase means output is synchronised with the input ) goes off the scale in the negative direction. My bet is right down to 180 degrees and thus indicating the push opposes the motion. So when the earth moves it grips harder while opposing the motion! As it should, because pushing in the same direction the mirror is already moving would amplify the problem. The maths simply describe the detail of precisely how much and when. One can alter the response characteristics to suit the task.
Note that at 30 odd Hz it's only some electromechanical loop that could respond fast enough to damp any induced vibration. But you also want alot of freedom to align the device correctly during the time that one is acquiring lock. So there is one filter used for alignment ( the weaker ) to allow adjustment to occur. But when that is settled, the interferometer is locked and taking valid data, you want to hold tight on that configuration ie. 'strong man mode while locked'. So you add in the other stronger filter, and the total response is shown as the red/brown curve. ( While mentioned in the legend, I don't see plotted the fourth curve referring to the label 'original stop' ).
It looks like they'll be experimenting with this for a while. But it's an important issue as the WFS yaw circuit has been identified as one which needed correction. Particularly with seismic disturbances. It's notable that it is the yaw mode that needs correction. Most earth tremors rock you side to side, with a somewhat lesser up/down movement.
Cheers, Mike.
( edit ) Here's an animation from an earlier thread illustrating the ( very exaggerated ) movement axes and modes :
Not all optical elements have all these degrees of freedom, but it shows how to describe them.
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
HAM6 is go for close up
We've done the following this morning:
- Leveled the HAM6 ISI, and left it unlocked
- Geophones show good free ~1Hz motion as expected
- Realigned the TTs to get ~90 cts on QPDs 1,2,3,4 (biases are saved)
- Double checked all earthquake stops were free from bread board and penultimate mass, and then lock-nutted.
- Double checked that everything is bolted down, and no tools were left on the table
- OMC SUS is free (see Tobin's entry for spectra)
I chuckle at the 'no tools' comment, it's an issue that has bitten probably many generations of technicians. Here's a neat diagram :
I think the beam comes in at the top of the page from the arm, is bounced to the far left then back over to the right. It goes through some cat's cradle of a path which I think is the output mode cleaner. There are a number of 'pick off' points in amongst this arrangement, where the light flux is converted to an electrical current by various photodiodes ( marked with the characters 'PD' within their label ).
And a view looking down on one of the mirrors, I think 'TT0' at about the table's centre :
The joy of splitting mirrors is that you get some of the light reflecting and some transmitting through. To a certain extent all mirrors do this, it's a matter of degree. You don't get something for nothing though as the intensity will diminish with each mirror encounter. Also there are phase change issues to keep track of. What is of especial interest is the V shaped gadget below the angled mirror in the middle of the picture. I reckon it's a beam dump, meaning we want to absorb any light transmitting through the mirror and not left to bounce about any old how in the apparatus. Looking closely at the inner side walls of the V it is reflecting the metal surface it sits on :
Thus they are mirrors too. But if light enters from any direction via the 'open jaw' of the V then it can only be reflected ( angle of incidence = angle of reflection ) progressively further into the narrow closed end, where it is finally absorbed.
The other thing is all that shiny metal framework and whatnot. What do ya reckon, Invar or something with a nice low thermal expansion co-efficient?
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
At Hanford they are pumped down and back to some degree of science operation. They achieved good lock at 14W laser power, yielding a peak of 17.9 Mpc range!! I think that might be a record of sorts. However a seismic event knocked them out and they couldn't get it back to lock beyond about 8W. It seems the WFS control is the problem, in pitch apparently -
Quote:
I continued to have problems getting a decent alignment and making it past Detect. I finally traced the problem to IOOWFS1 pitch, which was consistently walking away with the beam.
At Livingston they are yet to get to vacuum, because of a pump problem. There is also an issue with the Hydraulic External Pre-Isolation ( HEPI ). This is an active system ( of boxes within boxes now ) that serve to reduce the effect of the ground shaking upon the beam. The ground at Livingston is known to be more 'wobbly' than at Hanford. For a given seismic disturbance the soil vibrates about seven times more.
Cheers, Mike.
( edit ) It ought not but one wonders whether the new yaw filter may be relevant.
( edit ) WFS1 is the sensor at the dark port. Trouble is that it is sensitive ( will change value ) to changes of pointing at many mirrors.
( edit ) At Livingston, that extra wobbly ground plus the fact that alot more goes on in the area surrounding the site ( commercial forest ) was the reason for poor duty cycles in earlier runs. Hence the earlier than planned introduction of the more advanced seismic isolation systems, which I believe were originally slated to be installed as part of Advanced LIGO.
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
In an idle moment or three, following up on the BNC business, I browsed through various LIGO design documents and came across this one. It sounds nerdy, but I was impressed by the degree of specification there-in. The idea is to regularise the hardware used to a standard so that usage, failures and fault tracking can be managed better. The document discusses electrical interfaces - the physical connections to transmit power and signals etc. It's this sort of low level detail which could be crucial to success though so I don't think it should be dismissed as minor detail. We are looking for effects 21 orders of magnitude below human scale.
For signals the recommendation is to use D-sub type connectors :
You'll recognise this from the back of your computer case of course. Note the screw thread to achieve 'strain relief', meaning the pins don't partake in the forces to keep the junction together. Plus one should have the bolt which winds into this not falling off when it is not threading the hole.
If there's alot of connections to make across a single plug then SCSI type is preferred ( again with strain relief ) :
As for power connectors the idea is to have the female side as active - just like the wall plugs at home - to reduce risk of shorts. And suitable insulation at endings depending on voltage to be handled. And here is the backshell arrangement :
Plus there's detail relating to the flexible type of connectors that you tend to see within laptops, say.
What I guess this comes down to isn't simply correct and safe operation of components and systems but also the issue discussed earlier of signal leaks, resonances and whatnot.
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
On my SUN WS I have no serial connector, no parallel connector. I have 6 USB connectors, 2 FireWire connectors, 2 Ethernet connectors, 5 audio and micro connectors, plus the mains connector for my 220V AC.Plus a DB15 VGA connector.
Tullio
At Hanford they are having a lot of trouble with the Output Mode Cleaner, which I will comment upon later if & when I can make sense of it. I thought I'd talk of something simpler :
Quote:
Helicopter Flybys
The control room got a call today about more helicopter surveys. The guy on the other end said they will be flying around tomorrow, for a few hours in the mid-morning and a few hours in the mid-afternoon. He also said they may be flying around on Sunday as well.
You may ask, so what? What is the influence of said helicopters?
The obvious, and correct, answer is vibration. Anyone who has stood nearby one when operating can feel the buffeting from the 'blade wash'. What gives the chop-chop sound is ( generally ) the tips of the blades exceeding the speed of sound. One can hear this come into play as they power up to full and the rotors spin faster. This produces a shock wave, a mini sonic boom, which is a sharp increase in air pressure over a short distance. The 'crack' of a .303 round is similar as the bullet exceeds the sound barrier. Now the energy carried away from the source is going to hit against the sides of the containing buildings, plus cause ground vibration which also transmits into the interferometer. This will be sensed as frequencies related to the source, harmonics etc ...
What is a not so obvious effect is what is referred to as 'gravity gradient noise', a somewhat obtuse terminology that refers to movements of masses. One normally doesn't consider this of course in everyday life, but strictly speaking when the helicopter flies by there will be a local change in the force of gravity. The helicopter has mass, is subject to Newton's law of gravitation, so the component of gravitational force you feel due to the helicopter's presence will alter as it moves. Though because G is so small, it's not like you will lean towards the helicopter as it goes past though! :-)
The term 'gradient' is a mathematical reference to how potential energy is related to force strength and direction. Basically if you express matters in a field theory then the maximum rate of change of the potential indicates the line/direction of the force. This is a multi-dimensional comment, so in three dimensions you need a 3-D derivative operation called, not surprisingly, the gradient of the field. So you derive a vector field ( an arrow at each point in space ) from a scalar field ( a single number at each point in space ) by examining which infinitesimal step away from a given point gives the greatest positive change in the scalar value. [ Actually by convention the force is in the opposite direction to that maximum, towards regions of lower potential ]. This is very analogous to 'water flows downhill to seek the lowest point', or the 'fall line' on a ski slope which gives you the fastest ride! :-)
In any case if you move objects with mass around then there will be, however small, an effect on other objects. Including the interferometer. Because such objects around our interferometers move relatively slowly ( compared to light ) then this can be understood without either Relativity ie. Newton. That also means that changes to the local gravity field will be quite slow, and hence are low frequency signals. This is why the left end of the noise budget curves we have examined earlier show a significant rise. Clearly :
F = G * M1 * M2 / R^2
indicates a greater effect with bigger masses and closer proximity. But even a small bird flying right over the interferometer could have a bigger influence than say a large truck driving miles away because of the inverse square of the distance. In any case when you move a mass you change the scalar potential values, thus the local gradients change and hence the term 'gravity gradient noise'. But it's another way of stating our understanding of forces.
If you want to operate an interferometer here on Earth, you can't escape this. What would you do? Exclude activity for miles around, airspace also? Operate by remote control without mobile humans on the property? Yup, you guessed it, you go into space ......
Cheers, Mike.
( edit ) Here's an interesting example to illustrate the idea of potential. Consider a 'test' mass, say me in a spacesuit, floating in the near space around Earth. I say 'test' because my presence is not going to significantly alter the Sun or the Earth. Ignore the Moon for the moment. One can calculate the potential energy I have due to the presence of the Earth, and also due to the Sun. My total is a simple sum of the two ( this is a nice thing about such 'conservative' fields ). If I move about the place closer to and further from the Earth I can compare which potential term ( Earth or Sun ) is the greater. And I can ask the question 'when are they equal?' It turns out that at about 40 Earth radii they are pretty much the same ( exact distance depends on what side of the Earth I am on - side nearest the Sun or whatever ).
Now here is the surprise. The Moon is at around 60 Earth radii from Earth. Hence the Sun is the dominant effect on the Moon and not the Earth. You can legitimately say that the Moon orbits around the Sun, with a lesser perturbation from the presence of the Earth! So while the Sun is quite a long way away, it is absolutely enormous and sufficiently so to negate the effect of the distance inverse square character.
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
Down Unda we have a bull they call 'Chainsaw' ( there's been several by that name actually ), that is used in the top rodeos. Not quite a widow-maker but has potential. A whole year can go by with no-one managing to stay on the standard 10 seconds! The Hanford interferometer seems like that at present, throwing the operators off very readily. :-(
There's been a host of influences : the Output Mode Cleaner, the Thermal Compensation System and also the table on which the optics sit ( LVEA - ISCT4 ). It sits on a cushion of air, so to speak, in order to reduce vibration. But they found it was less noisy to have it 'landed' - the air pressure reduced to nil in the relevant supports - than floating.
When the laser power is increased, lock is lost. One operator managed to step gradually up :
Quote:
I took H1 to 8W powerup with usual scripts.
Ran powerUpIFO.pl to 11,12,13,14W in immediate succession.
Remained at 14W ~45 min
Increased power to 15W.
Remained at 15W ~10 min
Increased power to 16W.
Remained at 16W ~10 min
Increased power to 17W.
Lost lock after ~7min ( no obvious seismic/ environmental cause for lockloss)
So what gives? That's under investigation .....
However the above transitions were made in RF mode, without the output mode cleaner, in the fashion of previous science runs. Livingston has had much better uptime at a modest range.
And here's the gossip on the choppers from the TriCity Herald, Friday, Sept. 25, 2009 :
Caption reads "CH2M Hill Plateau Remediation Co. flies a helicopter low over the Hanford shrub steppe to detect where animals may have spread radioactive waste."
Let's not chuckle, as I did initially. It's Plutonium.
Cheers, Mike.
( edit ) Sorry, I should have included a link to the full story on the waste.
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
Hanford is the site where the plutonium for the Trinity text explosion and the Nagasaki bomb was produced by a nuclear reactor. There is a huge and costly cleaning and decommissioning operation going on.
Tullio
Bad vibrations aside, did the interferometer gravitationally 'see' the helicopter?
Yes, probably. But not in an immediately appreciable sense. What effect the chopper has will enter into all the other effects on the interferometer arms. To attribute a particular helicopter flyby, say, would require pretty well what we are doing here at E@H. Signal analysis. Trying to find that effect in amongst all the others. Like attempting to hear one 'off' note from a single instrument within a large orchestra, and while they are playing the 1812 Overture! Or Mamma Mia. :-)
So you can see why more than one interferometer, and separated by decent distances, is the desired method. So a chopper in Washington State isn't going to influence LIGO in Louisiana. Unless it flies down there, in which case that disturbance won't be near simultaneous, helicopters being sub-light speed vehicles. Ditto for a tree falling in the forest at Livingston affecting Hanford. Actually there's a neat spin on the 'if a tree falls in a forest when no-one is there, does it make a sound?' philosophical gag. Well the 2009 answer is 'yes, Einstein at Home hears it!' :-)
Quote:
(Thanks for the nicely illuminating commentary!)
You're welcome! I hope I don't make too many errors or mislead. I find it forces me to read stuff and think about the project more. I hope people aren't too shy to ask questions. The view count for this thread indicates I have a bit of an audience for this, so I want to stay sharp and not blow it! :-) :-)
Interestingly Kip Thorne ( Caltech, one of the 'designers' of LIGO ) says he learnt more physics when considering LIGO than anything else he's done. Which is a significant comment from someone of his stature. I guess the act of trying to catch such subtle signals, with measurement situations near quantum limits, really hones the understanding.
I do have a correction actually. As regards shot noise. What I explained earlier only really refers to photon generation by the laser. In fact that is a ( gravitational wave ) frequency independent effect, and so doesn't explain the higher frequency problem. What also happens is spontaneous emission by the laser, in addition to stimulated emission. Basically if an atom in the laser tube is in a state where it could emit a photon, that might happen anyway regardless of any passing photon of the same energy/frequency. That's why laser is an acronym - Light Amplification by Stimulated Emission of Radiation. It's because photons are boson type particles. They like to hang around together in the same state, unlike fermions ( eg. electrons, protons, neutrons .... ) that obey the Pauli Exclusion Principle and avoid each other. So it's sort of like a passing photon likes to pick up an identical hitch-hiker on the way through. Which is why photons produced by the laser process have the same phase, as the bosonic behaviour of matching photon states includes the phase. But even if no lasing is happening you'd still get some photons emitted, but not coherent in phase now. So that means further down the line in the interferometer one will get essentially random arrival of photons from such spontaneous emission. That still doesn't get us the higher noise on the right end of the energy budget curve though.
[ For the mathematically inclined, the arrivals have a Poisson distribution - which models very well ( apparently ) incoherent/unrelated things like say lightning strikes, or cars arriving at the McDonald's drive-through queue. ]
What does give the shot noise curve rising at the higher gravitational wave frequencies is a bit more subtle. Again. Recall that higher accuracy in frequency requires longer time intervals of measurement. So the photons that give us the high frequency information have been in the interferometer longer. They have oscillated back and forth up/down the arms more times. Even though the 4km arm length is generally much smaller than the wavelength of the gravitational waves we are trying to detect, it still means that those persistent photons could well experience a significant shift in the position of the cavity mirrors while they are circulating. Note that higher frequency gravitational waves will have a shorter wavelength as well. For instance when arriving at the X-arm end during one lap : the mirror there may have moved ( from it's 'neutral' position ) away from the corner station. But a few laps later that same mirror may now be closer to the corner station. So any photons around in the arm during that interval will experience consecutive bounces from the mirrors when they are in a different positions. This effectively 'blurs' what we are trying to measure, because the inter-mirror distance has altered while we were doing it. So the arrival times of the photons at our photodetectors will vary accordingly. We left the camera shutter open too long, so to speak.
This stuff is really the guts of the interferometer method. The passing gravitational wave does really affect matter and radiation quite differently. If it didn't, we'd have no hope.
Cheers, Mike.
( edit ) Actually it'll be more than 'a few laps later'. My back of the envelope estimate, with speed of light @ 300,000 km /sec :
At 750 Hz ( gravitational ) wavelength = 300000 / 750 = 400 km
50 circuits x ( 2 X 4km ) per circuit = 400 km
So those photons would be in the interferometer for a bit more than a millisecond. Enough time for an appreciable fraction of the waveform to pass by.
Same principle though.
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
From Hanford : RE: 0.
)
From Hanford :
The Wave Front Sensors are used to make sure that we are looking at the right parts of the light coming through the dark port of the interferometer. We want the face of the photodiode to be 'square on' with respect to any light coming through. Or equivalently we want any/all the photons we catch to be in the same phase. So if the photodiode is 'skewed' then we will be receiving photons that have a spread of phase. If the interferometer is otherwise properly locked and whatnot then we would get some photons that have returned precisely one half wavelength ( or odd multiple ) out of phase with respect to the other arm, some a bit less than half a wavelength, and some a bit more than half a wavelength. This effectively introduces an unnecessary length uncertainty, and behaves like a depth of field effect in camera optics. This is the 'angle to length coupling' mentioned above. We want to 'focus' as best as possible on an exact plane.
Michelson and Morley did this deliberately. They didn't have sensitive photodiodes alas! So they arranged matters so that they could view a spread of phases. It appeared as fringes, a series of light and dark bands. However what they were looking for was a shift in the fringe pattern, which they had some hope of detecting with The Mark I Eyeball Instrument, as the whole interferometer was rotated. Thus possibly demonstrating a directional dependence to light transmission, aether etc .....
Thus the WFS must participate in the adjustment of the interferometer, which is an active beast that continually self-adjusts via various control loops. One of these feedback mechanisms is the yaw filter on the WFS. Yaw is an angular deflection. In an aeroplane it is a left/right rotation about an axis through the centre of gravity and also perpendicular to both the wing axis and the fuselage axis. Or, say, if you look straight ahead and then rotate your neck to look left/right without nodding that's also yaw ( looking up/down is designated as pitch, tilting to left/right without nodding is roll ). Filter is the answer to the question - if the mirror is disturbed in yaw then what correction do I apply to the servos that hold the mirror aligned? The servos are little pushing/pulling devices to point the mirror.
There are two features to the correction : magnitude and phase. The magnitude is how much do the servos push. The phase refers to response to an oscillation, basically do you push in the same or opposite direction to the oscillation? ( this is not the photon phase discussed above ). Anyone who has had fun on a playground swing already implicitly knows all about this. It's like a pendulum. You can have someone push you of course, many do. But you can also 'pump up' the motion yourself by raising and lowering your center of mass with respect to the seat. Either way if the timing is right then the swing will increase, but you can time differently and suppress the motion too. As a kid I would spend heaps of time stuffing about with this aspect of swings - and quite naturally without any physics tuition! I bet you have too .... :-)
You can think of the filter as a black box with an input and an output. The input is how the mirror is deviating from some 'neutral' position, and the output is the magnitude/phase as above. You describe the behaviour of the filter by plotting what output happens for a given input :
Don't panic when you look at this. Basically for a given frequency of disturbance you look along the horizontal axis ( which is logarithmic to compress the display ) to find it and then go up to hit one of the colored curves. Then go from that point on the curve across to the left side axis and read off the value. The interesting area is the 30/50-ish Hz region.
The green curve is what the filter had been, which was giving trouble. Essentially when a jiggle would come along in this frequency range then the magnitude of the response would go down. The left scale for magnitude is marked in decibels. It's another logarithmic thing. Zero dB means no change in strength, positive ( up the axis ) is stronger, negative ( down the axis ) is weaker. So when things started rocking ( an earth tremor say ) the grip would relax, the oscillation would worsen and probably cause the grip to relax some more.
The blue curve is a new filter. It goes up in that frequency band and hence holds on harder. Plus if you note the phase curve ( where zero phase means output is synchronised with the input ) goes off the scale in the negative direction. My bet is right down to 180 degrees and thus indicating the push opposes the motion. So when the earth moves it grips harder while opposing the motion! As it should, because pushing in the same direction the mirror is already moving would amplify the problem. The maths simply describe the detail of precisely how much and when. One can alter the response characteristics to suit the task.
Note that at 30 odd Hz it's only some electromechanical loop that could respond fast enough to damp any induced vibration. But you also want alot of freedom to align the device correctly during the time that one is acquiring lock. So there is one filter used for alignment ( the weaker ) to allow adjustment to occur. But when that is settled, the interferometer is locked and taking valid data, you want to hold tight on that configuration ie. 'strong man mode while locked'. So you add in the other stronger filter, and the total response is shown as the red/brown curve. ( While mentioned in the legend, I don't see plotted the fourth curve referring to the label 'original stop' ).
It looks like they'll be experimenting with this for a while. But it's an important issue as the WFS yaw circuit has been identified as one which needed correction. Particularly with seismic disturbances. It's notable that it is the yaw mode that needs correction. Most earth tremors rock you side to side, with a somewhat lesser up/down movement.
Cheers, Mike.
( edit ) Here's an animation from an earlier thread illustrating the ( very exaggerated ) movement axes and modes :
Not all optical elements have all these degrees of freedom, but it shows how to describe them.
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
From Livingston, they are
)
From Livingston, they are going back to vacuum :
I chuckle at the 'no tools' comment, it's an issue that has bitten probably many generations of technicians. Here's a neat diagram :
I think the beam comes in at the top of the page from the arm, is bounced to the far left then back over to the right. It goes through some cat's cradle of a path which I think is the output mode cleaner. There are a number of 'pick off' points in amongst this arrangement, where the light flux is converted to an electrical current by various photodiodes ( marked with the characters 'PD' within their label ).
And a view looking down on one of the mirrors, I think 'TT0' at about the table's centre :
The joy of splitting mirrors is that you get some of the light reflecting and some transmitting through. To a certain extent all mirrors do this, it's a matter of degree. You don't get something for nothing though as the intensity will diminish with each mirror encounter. Also there are phase change issues to keep track of. What is of especial interest is the V shaped gadget below the angled mirror in the middle of the picture. I reckon it's a beam dump, meaning we want to absorb any light transmitting through the mirror and not left to bounce about any old how in the apparatus. Looking closely at the inner side walls of the V it is reflecting the metal surface it sits on :
Thus they are mirrors too. But if light enters from any direction via the 'open jaw' of the V then it can only be reflected ( angle of incidence = angle of reflection ) progressively further into the narrow closed end, where it is finally absorbed.
The other thing is all that shiny metal framework and whatnot. What do ya reckon, Invar or something with a nice low thermal expansion co-efficient?
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
At Hanford they are pumped
)
At Hanford they are pumped down and back to some degree of science operation. They achieved good lock at 14W laser power, yielding a peak of 17.9 Mpc range!! I think that might be a record of sorts. However a seismic event knocked them out and they couldn't get it back to lock beyond about 8W. It seems the WFS control is the problem, in pitch apparently -
At Livingston they are yet to get to vacuum, because of a pump problem. There is also an issue with the Hydraulic External Pre-Isolation ( HEPI ). This is an active system ( of boxes within boxes now ) that serve to reduce the effect of the ground shaking upon the beam. The ground at Livingston is known to be more 'wobbly' than at Hanford. For a given seismic disturbance the soil vibrates about seven times more.
Cheers, Mike.
( edit ) It ought not but one wonders whether the new yaw filter may be relevant.
( edit ) WFS1 is the sensor at the dark port. Trouble is that it is sensitive ( will change value ) to changes of pointing at many mirrors.
( edit ) At Livingston, that extra wobbly ground plus the fact that alot more goes on in the area surrounding the site ( commercial forest ) was the reason for poor duty cycles in earlier runs. Hence the earlier than planned introduction of the more advanced seismic isolation systems, which I believe were originally slated to be installed as part of Advanced LIGO.
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
In an idle moment or three,
)
In an idle moment or three, following up on the BNC business, I browsed through various LIGO design documents and came across this one. It sounds nerdy, but I was impressed by the degree of specification there-in. The idea is to regularise the hardware used to a standard so that usage, failures and fault tracking can be managed better. The document discusses electrical interfaces - the physical connections to transmit power and signals etc. It's this sort of low level detail which could be crucial to success though so I don't think it should be dismissed as minor detail. We are looking for effects 21 orders of magnitude below human scale.
For signals the recommendation is to use D-sub type connectors :
You'll recognise this from the back of your computer case of course. Note the screw thread to achieve 'strain relief', meaning the pins don't partake in the forces to keep the junction together. Plus one should have the bolt which winds into this not falling off when it is not threading the hole.
If there's alot of connections to make across a single plug then SCSI type is preferred ( again with strain relief ) :
As for power connectors the idea is to have the female side as active - just like the wall plugs at home - to reduce risk of shorts. And suitable insulation at endings depending on voltage to be handled. And here is the backshell arrangement :
Plus there's detail relating to the flexible type of connectors that you tend to see within laptops, say.
What I guess this comes down to isn't simply correct and safe operation of components and systems but also the issue discussed earlier of signal leaks, resonances and whatnot.
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
On my SUN WS I have no serial
)
On my SUN WS I have no serial connector, no parallel connector. I have 6 USB connectors, 2 FireWire connectors, 2 Ethernet connectors, 5 audio and micro connectors, plus the mains connector for my 220V AC.Plus a DB15 VGA connector.
Tullio
At Hanford they are having a
)
At Hanford they are having a lot of trouble with the Output Mode Cleaner, which I will comment upon later if & when I can make sense of it. I thought I'd talk of something simpler :
You may ask, so what? What is the influence of said helicopters?
The obvious, and correct, answer is vibration. Anyone who has stood nearby one when operating can feel the buffeting from the 'blade wash'. What gives the chop-chop sound is ( generally ) the tips of the blades exceeding the speed of sound. One can hear this come into play as they power up to full and the rotors spin faster. This produces a shock wave, a mini sonic boom, which is a sharp increase in air pressure over a short distance. The 'crack' of a .303 round is similar as the bullet exceeds the sound barrier. Now the energy carried away from the source is going to hit against the sides of the containing buildings, plus cause ground vibration which also transmits into the interferometer. This will be sensed as frequencies related to the source, harmonics etc ...
What is a not so obvious effect is what is referred to as 'gravity gradient noise', a somewhat obtuse terminology that refers to movements of masses. One normally doesn't consider this of course in everyday life, but strictly speaking when the helicopter flies by there will be a local change in the force of gravity. The helicopter has mass, is subject to Newton's law of gravitation, so the component of gravitational force you feel due to the helicopter's presence will alter as it moves. Though because G is so small, it's not like you will lean towards the helicopter as it goes past though! :-)
The term 'gradient' is a mathematical reference to how potential energy is related to force strength and direction. Basically if you express matters in a field theory then the maximum rate of change of the potential indicates the line/direction of the force. This is a multi-dimensional comment, so in three dimensions you need a 3-D derivative operation called, not surprisingly, the gradient of the field. So you derive a vector field ( an arrow at each point in space ) from a scalar field ( a single number at each point in space ) by examining which infinitesimal step away from a given point gives the greatest positive change in the scalar value. [ Actually by convention the force is in the opposite direction to that maximum, towards regions of lower potential ]. This is very analogous to 'water flows downhill to seek the lowest point', or the 'fall line' on a ski slope which gives you the fastest ride! :-)
In any case if you move objects with mass around then there will be, however small, an effect on other objects. Including the interferometer. Because such objects around our interferometers move relatively slowly ( compared to light ) then this can be understood without either Relativity ie. Newton. That also means that changes to the local gravity field will be quite slow, and hence are low frequency signals. This is why the left end of the noise budget curves we have examined earlier show a significant rise. Clearly :
F = G * M1 * M2 / R^2
indicates a greater effect with bigger masses and closer proximity. But even a small bird flying right over the interferometer could have a bigger influence than say a large truck driving miles away because of the inverse square of the distance. In any case when you move a mass you change the scalar potential values, thus the local gradients change and hence the term 'gravity gradient noise'. But it's another way of stating our understanding of forces.
If you want to operate an interferometer here on Earth, you can't escape this. What would you do? Exclude activity for miles around, airspace also? Operate by remote control without mobile humans on the property? Yup, you guessed it, you go into space ......
Cheers, Mike.
( edit ) Here's an interesting example to illustrate the idea of potential. Consider a 'test' mass, say me in a spacesuit, floating in the near space around Earth. I say 'test' because my presence is not going to significantly alter the Sun or the Earth. Ignore the Moon for the moment. One can calculate the potential energy I have due to the presence of the Earth, and also due to the Sun. My total is a simple sum of the two ( this is a nice thing about such 'conservative' fields ). If I move about the place closer to and further from the Earth I can compare which potential term ( Earth or Sun ) is the greater. And I can ask the question 'when are they equal?' It turns out that at about 40 Earth radii they are pretty much the same ( exact distance depends on what side of the Earth I am on - side nearest the Sun or whatever ).
Now here is the surprise. The Moon is at around 60 Earth radii from Earth. Hence the Sun is the dominant effect on the Moon and not the Earth. You can legitimately say that the Moon orbits around the Sun, with a lesser perturbation from the presence of the Earth! So while the Sun is quite a long way away, it is absolutely enormous and sufficiently so to negate the effect of the distance inverse square character.
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
Down Unda we have a bull they
)
Down Unda we have a bull they call 'Chainsaw' ( there's been several by that name actually ), that is used in the top rodeos. Not quite a widow-maker but has potential. A whole year can go by with no-one managing to stay on the standard 10 seconds! The Hanford interferometer seems like that at present, throwing the operators off very readily. :-(
There's been a host of influences : the Output Mode Cleaner, the Thermal Compensation System and also the table on which the optics sit ( LVEA - ISCT4 ). It sits on a cushion of air, so to speak, in order to reduce vibration. But they found it was less noisy to have it 'landed' - the air pressure reduced to nil in the relevant supports - than floating.
When the laser power is increased, lock is lost. One operator managed to step gradually up :
So what gives? That's under investigation .....
However the above transitions were made in RF mode, without the output mode cleaner, in the fashion of previous science runs. Livingston has had much better uptime at a modest range.
And here's the gossip on the choppers from the TriCity Herald, Friday, Sept. 25, 2009 :
Caption reads "CH2M Hill Plateau Remediation Co. flies a helicopter low over the Hanford shrub steppe to detect where animals may have spread radioactive waste."
Let's not chuckle, as I did initially. It's Plutonium.
Cheers, Mike.
( edit ) Sorry, I should have included a link to the full story on the waste.
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
Hanford is the site where the
)
Hanford is the site where the plutonium for the Trinity text explosion and the Nagasaki bomb was produced by a nuclear reactor. There is a huge and costly cleaning and decommissioning operation going on.
Tullio
RE: ... What is the
)
So...
Bad vibrations aside, did the interferometer gravitationally 'see' the helicopter?
Regards,
Martin
(Thanks for the nicely illuminating commentary!)
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
RE: Bad vibrations aside,
)
Yes, probably. But not in an immediately appreciable sense. What effect the chopper has will enter into all the other effects on the interferometer arms. To attribute a particular helicopter flyby, say, would require pretty well what we are doing here at E@H. Signal analysis. Trying to find that effect in amongst all the others. Like attempting to hear one 'off' note from a single instrument within a large orchestra, and while they are playing the 1812 Overture! Or Mamma Mia. :-)
So you can see why more than one interferometer, and separated by decent distances, is the desired method. So a chopper in Washington State isn't going to influence LIGO in Louisiana. Unless it flies down there, in which case that disturbance won't be near simultaneous, helicopters being sub-light speed vehicles. Ditto for a tree falling in the forest at Livingston affecting Hanford. Actually there's a neat spin on the 'if a tree falls in a forest when no-one is there, does it make a sound?' philosophical gag. Well the 2009 answer is 'yes, Einstein at Home hears it!' :-)
You're welcome! I hope I don't make too many errors or mislead. I find it forces me to read stuff and think about the project more. I hope people aren't too shy to ask questions. The view count for this thread indicates I have a bit of an audience for this, so I want to stay sharp and not blow it! :-) :-)
Interestingly Kip Thorne ( Caltech, one of the 'designers' of LIGO ) says he learnt more physics when considering LIGO than anything else he's done. Which is a significant comment from someone of his stature. I guess the act of trying to catch such subtle signals, with measurement situations near quantum limits, really hones the understanding.
I do have a correction actually. As regards shot noise. What I explained earlier only really refers to photon generation by the laser. In fact that is a ( gravitational wave ) frequency independent effect, and so doesn't explain the higher frequency problem. What also happens is spontaneous emission by the laser, in addition to stimulated emission. Basically if an atom in the laser tube is in a state where it could emit a photon, that might happen anyway regardless of any passing photon of the same energy/frequency. That's why laser is an acronym - Light Amplification by Stimulated Emission of Radiation. It's because photons are boson type particles. They like to hang around together in the same state, unlike fermions ( eg. electrons, protons, neutrons .... ) that obey the Pauli Exclusion Principle and avoid each other. So it's sort of like a passing photon likes to pick up an identical hitch-hiker on the way through. Which is why photons produced by the laser process have the same phase, as the bosonic behaviour of matching photon states includes the phase. But even if no lasing is happening you'd still get some photons emitted, but not coherent in phase now. So that means further down the line in the interferometer one will get essentially random arrival of photons from such spontaneous emission. That still doesn't get us the higher noise on the right end of the energy budget curve though.
[ For the mathematically inclined, the arrivals have a Poisson distribution - which models very well ( apparently ) incoherent/unrelated things like say lightning strikes, or cars arriving at the McDonald's drive-through queue. ]
What does give the shot noise curve rising at the higher gravitational wave frequencies is a bit more subtle. Again. Recall that higher accuracy in frequency requires longer time intervals of measurement. So the photons that give us the high frequency information have been in the interferometer longer. They have oscillated back and forth up/down the arms more times. Even though the 4km arm length is generally much smaller than the wavelength of the gravitational waves we are trying to detect, it still means that those persistent photons could well experience a significant shift in the position of the cavity mirrors while they are circulating. Note that higher frequency gravitational waves will have a shorter wavelength as well. For instance when arriving at the X-arm end during one lap : the mirror there may have moved ( from it's 'neutral' position ) away from the corner station. But a few laps later that same mirror may now be closer to the corner station. So any photons around in the arm during that interval will experience consecutive bounces from the mirrors when they are in a different positions. This effectively 'blurs' what we are trying to measure, because the inter-mirror distance has altered while we were doing it. So the arrival times of the photons at our photodetectors will vary accordingly. We left the camera shutter open too long, so to speak.
This stuff is really the guts of the interferometer method. The passing gravitational wave does really affect matter and radiation quite differently. If it didn't, we'd have no hope.
Cheers, Mike.
( edit ) Actually it'll be more than 'a few laps later'. My back of the envelope estimate, with speed of light @ 300,000 km /sec :
At 750 Hz ( gravitational ) wavelength = 300000 / 750 = 400 km
50 circuits x ( 2 X 4km ) per circuit = 400 km
So those photons would be in the interferometer for a bit more than a millisecond. Enough time for an appreciable fraction of the waveform to pass by.
Same principle though.
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