Detector Watch

Chipper Q
Chipper Q
Joined: 20 Feb 05
Posts: 1540
Credit: 708571
RAC: 0

RE: You may be wondering

Quote:
You may be wondering why they don't just go in and casually manhandle the mirror etc. It takes ages to realign and re-settle the whole array - so that may be a backward step, not to mention the risk of introducing some new error.


Very good point. As they continue to look closely at the suspension for Hanford's optics, we have a chance to see some plots that aren't generated often. The data comes from rare times when the all the optics are undamped, referred to as 'Free Swing Data'. Here is a log entry from Oct. 25 (Hanford):

Quote:

Thus far in S5, we've had a couple of periods where we (1)had ALL optics undamped for more than 45min, and (2)had no FB issues (so, power outtages, don't count!). The dates in question are:

5/20/06--Thunderstorm
8/29/06--Tues Maintenance work (power distrib work)
Oct------Several times to pick from

So, with three data points, wanted to see how each optic has been looking. Now, grabbing some of the old data is possible, but it can take a while. So far, I've looked at H1's ETMs, ITMs, & BS. Will continue to look at ALL optics during upcoming shifts.

Now for these measurements, the main thing we want to look at are the red traces. These are the "current" traces (post Rainier quake). Want to see if any of these optics are culprits for our slighly lower operating Ranges on both ifos.


From comments associated with plots for a specific optic, the concern is with a shift in frequencies. But in this plot (for ITMX's free swing in the pitch-axis, I think), it looks like there's an extra spike that wasn't there before:
[img]http://groups.msn.com/_Secure/0SgDdAk0WK1ojv8cfC8!ZmCh8QKP2A5FVRQCM7SJMcMtjSpZblBA390tW6lg8f9rJUSCjuAUk68OUSuOFRY3VFoaU7M5rE6FBJ2zTiEoI9ZLf*M6tnuSnDA/FreeSwing.gif?dc=4675595055268435591[/img]
Are the peaks related to resonances in the suspension?
What are FB issues? (All I can think is 'forward bias'...)

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6534
Credit: 284734192
RAC: 103342

RE: Very good

Message 47957 in response to message 47956

Quote:
Very good point..............
Are the peaks related to resonances in the suspension?
What are FB issues? (All I can think is 'forward bias'...)


We certainly are seeing new data plots. I believe you are right in those changes on the frequency plot for pitch. Yes the peaks are due to resonances in the suspension. I suppose FB is 'forward bias' - I think this means an actuator signal which positions the mirror as part of keeping the ITM/ETM portion of the arm in ( Fabry-Perot ) resonance.

Now due to the magic of Macromedia Flash and my dear son Stephen's expertise with it ( I have no hope ) here are animations of the principal modes of oscillations of the mirrors:

Pitch: like nodding your head up and down

Yaw: like shaking your head from side to side

Roll: sort of like tilting your head from side to side. I'm not sure how much this motion applies to LIGO - I think the mounts absolutely constrain this.

Pendulum: like the playground swings we have all seen

Sideways pendulum: like the watch dangling used to hypnotise people

You can see the incoming laser from the left in purple, a reflected beam in green, the normal/perpendicular to the mirror centre in blue ( except for ROLL ). Reflection is specular here meaning angles of incidence and reflection are equal - assuming dead flat mirrors. The axis of rotation or line of swinging is in red.

All the movements are grossly exaggerated for this demonstration, in reality the isolation mechanisms damp things considerably. However bear in mind that the returning beam must go back up the pipe ( 2 to 4 km ) to hit the other mirror and return again etc. If a photon does many circuits during it's time in the system then small errors in positioning will accumulate with each passage. Quite an aiming job!

Each type of motion presents a different geometrical challenge for aiming.

Cheers, Mike.

( edit ) each animation has a different number of frames, and frame rates - so there's pauses etc. View them separately - Steve just whacked them into the one gif for upload.

( edit ) At any given time the actual motion of a mirror may be a combination/superposition of these modes ( with ones I haven't covered, plus harmonics et al ). You can see how examining these modes as separate concepts may help the LIGO people to tease out any problem behaviour.

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: 6534
Credit: 284734192
RAC: 103342

Sadly Hanford had to suffer

Sadly Hanford had to suffer the death of a hard disk - which happened to be a crucial one in their network:

Quote:

* All the monitor status reports were unavailable

* The LockLoss and CheckDataValid monitors died completely

* PlaneMon went into a loop

* The SPI page was braindead

As for Livingstone, the web page doesn't even load, but they are having a power outage of clearly considerable extent. L1 doesn't even appear on Hanford's State Vector plot! Oh dear.... what a month for Gremlins.

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

Chipper Q
Chipper Q
Joined: 20 Feb 05
Posts: 1540
Credit: 708571
RAC: 0

RE: ...due to the magic of

Quote:
...due to the magic of Macromedia Flash and my dear son Stephen's expertise with it...


Awesome job on the principal modes of oscillation animations! Not that I mind reading a thousand words, but good pictures convey the info so much faster!

Meanwhile, they're continuing to see some minor changes in alignment for some of the optics at Hanford. I noticed in one of the log entries (on the Graveyard Shift, Oct. 26) that burping also may have an effect on a particular mode for an optic. Indicative that maintaining the vacuum also presents measurable strain on the overall structure?

Not the best time for collecting science data: after yesterday's burp, pumps are running in the LVEA, the optics for H1 were in free swing mode all night, and H2 had trouble a couple times from seismic (including a 2.9mag quake at Mt St Helens), and also suffered lock loss once for reasons yet to be determined (it wasn't seismic). I've been looking at the spectrograms that were with the log entry. For 17 different sensors and 3 spectrograms per sensor, there's a lot to see! Here's a sample:
[img]http://groups.msn.com/_Secure/0SAAHA3EWOkerAIywuPlO1!3psKsM4PsLcmj!Gh*qjJ8svudXKCVGcXvqw65f9FnuU65z6E6bVBr8PtmbNulo!MGyFBv9DC6yl0fuMAAoqhRrAAAAynJxAg/Spectrogram.gif?dc=4675595236568072309[/img]
Cool visual aid, the spectrogram, neat way to picture the data! :)
- - -
edit: didn't get a chance to finish this post this morning... bad news indeed, about the /cvs disk! Awww.... looks like quick action switching to another drive "...got the IFOs through wed night..." but there's still lots to do before recovering completely from that :(

Chipper Q
Chipper Q
Joined: 20 Feb 05
Posts: 1540
Credit: 708571
RAC: 0

Still troubles at Livingston,

Still troubles at Livingston, besides recovering from a power outage on the 26th, and the troubles are mostly due to long periods of high microseismic activity, noted early (on the day shift of the 27th) and continuing into the evening shift, but that allowed some time to work on the mode cleaner, and perform checks on the PEM LVEA magnetometer spectra. This log entry from the morning summed it up pretty well for LLO:

Quote:

It looks like we can't get DMT data; none of the FOMs have updated since about two AM.

But the microseism is still too high to lock. In the attached plot, the reference
traces are from yesterday morning. The low band of the microseism has remained
hugely elevated, and the upper band has increased overnight.

Now is NOT a good time for a galactic supernova.

But it looks like Hanford is almost up to full steam, and hopefully they'll remain in Science Mode with both H1 and H2 during the 2.5 hour interval coming up shortly (07:00:00 - 09:30:00 UTC, on the 28th). Not because anyone's expecting a SN (or GRB {although SWIFT detected one at 01:26:22 UTC, 28/10}): RXTE is targeting Sco X-1, and they're determined to be ready for it at Hanford.

Found this quote in the story of Sco X1 (link above):

Quote:
I must admit that my main motivation for pressing forward was a deep seated faith in the boundless resourcefulness of nature, which so often leaves the most daring imagination of man far behind.
- Bruno Rossi
Chipper Q
Chipper Q
Joined: 20 Feb 05
Posts: 1540
Credit: 708571
RAC: 0

A quick look at the State

A quick look at the State Vector for the end of Owl shift at Hanford showed good news, and really good news, at a glance. Good news is Livingston made it back into Science Mode. Really good news is Hanford was in Science Mode (with both 2K and 4K arms) in time for the coincident observation of Sco X1 with the Rossi satellite! Looking through the various entries in the log, I found this (color added for emphasis):

Quote:

4K:
---------------
Locked most of the shift, while Keita worked on alignment sleuthing. The goldplate reflects many changes in alignment on the optics, dark offsets in the QPDs and WFS, etc.. Range averaged 14.5 MPc, touching 15.0 Mpc. Go Keita!
It's been in sm for about 1.5 hours.

2K:
---------------
Locked most of the shift, while Rana worked on it. Range averaged 6.7 Mpc, we'll be putting it into sm soon for some hopeful Sco X-1 coincidence.


and after a few more entries, I found this snippet:

Quote:
IFO is back in SM, for double coinc (no LLO yet) for the SCO X-1 monitoring.


Well done!

Was also fortunate enough to find an in-depth look at Mike's closer look on aligning the laser beam with all of the optics, courtesy of the Owl shift:

Quote:

[pre] Test Mass position offsets[/pre]
The F2A scripts balance out the coil gains using the optical levers and the
fact that driving the butterfly mode should produce no pit or yaw.

After running these scripts, we run the A2L scripts which adjust the pitch and
yaw output matrices of the suspensions to zero the coupling of angle drive to
length readout. With the coil gains balanced, these resulting pit/yaw coefficients
are only compensating for beam mis-centering on the test masses.

The optic diameter is 25 cm and the magnets are glued to within a couple mm of
the edge so that the distance between the UL and UR magnets is
25/sqrt(2) = 17.7 cm.

Some trigonometry soon shows that the spot decentering from center is:
[pre]
dia
Delta = ----------- * m
2 * sqrt(2)
[/pre]
where dia = 25 cm and m = 1 - abs(coef)

where coef is, e.g. H1:SUS-ITMY_ULYAW_GAIN.

so when one of our SUS output matrix coefficients is 1.01, that means that the
beam spot is 0.88 mm off center.


I recognize the trig as the familiar "1-1-sqrt(2) triangle", and dia/2 as the radius, but not sure if I'm applying it correctly to the mirror/magnet geometry... also, I had to use m = 10 * abs(1 – coef) to get the numbers in green, that I added to the this log entry (that followed the above explanation) showing actual values at the time, for the suspension system's output matrix coefficients:

Quote:

[pre] . Decentering (mm)
H1:SUS-ETMX_ULPIT_GAIN = 0.939 5.3 5.4
H1:SUS-ETMX_ULYAW_GAIN = 0.897 8.9 9.1

H1:SUS-ETMY_ULPIT_GAIN = 0.987 1.1 1.1
H1:SUS-ETMY_ULYAW_GAIN = 1.032 2.6 2.8

H1:SUS-ITMX_ULPIT_GAIN = 1.025 2.2 2.2
H1:SUS-ITMX_ULYAW_GAIN = 0.832 15.0 14.8

H1:SUS-ITMY_ULPIT_GAIN = 0.959 3.5 3.6
H1:SUS-ITMY_ULYAW_GAIN = 1.152 13.2 13.4

H2:SUS-ETMX_ULPIT_GAIN = 0.969 2.6 2.7
H2:SUS-ETMX_ULYAW_GAIN = 0.960 3.5 3.5

H2:SUS-ETMY_ULPIT_GAIN = 0.962 3.3 3.4
H2:SUS-ETMY_ULYAW_GAIN = 1.018 1.6 1.6

H2:SUS-ITMX_ULPIT_GAIN = 0.978 1.9 1.9
H2:SUS-ITMX_ULYAW_GAIN = 1.070 6.2 6.2

H2:SUS-ITMY_ULPIT_GAIN = 0.971 2.5 2.6
H2:SUS-ITMY_ULYAW_GAIN = 0.795 18.0 18.1[/pre]


Exciting to work with actual matrix values, and to see how the beam is controlled at each of the (many!) optical elements!

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6534
Credit: 284734192
RAC: 103342

RE: A quick look at the

Message 47964 in response to message 47963

Quote:
A quick look at the State Vector for the end of Owl shift at Hanford showed good news, and really good news, at a glance.......Exciting to work with actual matrix values, and to see how the beam is controlled at each of the (many!) optical elements!


Yeah, it's all happening at the Gabba, as they say...:-)

I think we need to spawn a new installment!

( I'll do this if the image/load size gets to big - to keep under 1 Meg or so )

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

Comment viewing options

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