| The LIGO
LIGO Hanford Observatory NewsSecond "First Lock" of the LIGO 2-km Interferometer
LIGO Discovers Moon and Sun!
The commissioning of the LIGO interferometers achieved another milestone this month. Last fall we achieved first lock of the LIGO 2-km interferometer, using precision sensing systems for the first time to hold all mirrors in exactly the right locations so that light would be bounced back and forth along the two arms in perfect synchrony. The light had to be brought to the beamsplitter precisely in phase to interfere distructively at the "dark port," and to interfere constructively at the "recycling mirror." This interferometer configuration is called a power-recycled Michelson interferometer with Fabry-Perot arms, and last fall was the first time LIGO--or any of the gravitational wave projects around the world--had achieved the necessary resonance conditions. The technical challenge was formidable, its magnitude can be understood by noting that the mirrors which form the interferometer had to be held in their correct positions within about 0.1 nm (the diameter of an atom!) even though they are separated by more than two kilometers. Needless to say, there was a real sense of accomplishment at achieving this feat. However, LIGO's first lock required one artificial element: we had to limit the build-up of light in the interferometer by causing some of the light to be lost. We could do this two ways: either by misaligning one of the mirrors (by less than an arcsecond!) so that the interference in the recycling cavity is not perfect, or by using one of the large valves on the vacuum system to skim off a small fraction (less than one percent) of the light in the arm cavity. Small as these imperfections might seem, they were enough to reduce the build-up of light in the interferometer (the so-called "recycling gain") by more than a factor of 30. The net result was an interferometer that was easier to control, and therefore easier to characterize and study, but one which would ultimately not be suitable for LIGO's purposes.
Since that time, scientists have been hard at work to improve the stability and tuning of the interferometer so as to achieve higher recycling gain. These efforts culminated in a dramatic week this month when the amount of time the interferometer could be held in lock with high recycling gain went from 0.1 second to more than 40 minutes. At times improvements seemed to come almost hourly. The current result, a recycling gain of 15, is by a factor of 2-3 less than had been expected, but further tuning of the system may bring that number closer to our expectations.
The figure at left above shows key images of the laser beam from four locations in the interferometer: on top of the two transmitted beams through the two arms, below, the "dark port," and a pickoff inside the recycling cavity. These are the same images as shown in the background of the first lock photo with Rai Weiss (above, right) but note that the saturated nature of the images indicates the increase in power in the interferometer (of order 500 for the transmitted beams).
Like many advances in experimental physics, the improvement in the locking performance was not the result of one change, but rather the sum of a large number of incremental changes, many worked out in the months prior to the final push:
-The servos which damp the angular modes of the suspended optics were switched from local sensors to optical levers.
-The light levels on photodetectors had to be optimized to give large enough signals, but not to saturate.
-The time delays between the light and the electrical signals had to be tuned to give the correct phase shifts to modulation signals.
-Electronic servo gains had to be adjusted dynamically as the interferometer came into resonance to compensate for the factor of several hundred higher light levels.
-The sensors which hold the interferometer in optimal alignment after lock is obtained had to be tuned up and brought into operation.
-The signals that are used by the control servos had to be understood over a wide range of configurations.
Each of these steps might require hours, sometimes days, of careful experimentation to pinpoint a problem, then the added time needed for hardware and software changes to implement a fix, and then still more hours of testing and understanding of the system after changes are made. Not to mention, of course, a few dead ends to meander down.
LIGO detected its first astronomical objects last November during a week-long engineering run at the LIGO Hanford Observatory!
Okay, so maybe the Moon and Sun were pretty well-known already. And no, we did not detect the spacetime curvature from these objects directly with our laser beams. And no, it won't pay the bills. But as a Tri-Cities Astronomy Club member pointed out, "It's like the first dollar bill taken in by a new business! Frame the data and hang it on the wall!"
The first week-long data record from the Hanford 2-km interferometer clearly showed the Earth Tide, the stretching and shrinking of Earth's crust under the gravitational forces from the Moon and the Sun. Roughly twice a day, the surface of the Earth's crust rises and falls a few centimeters. This motion is not easily detectable by LIGO, which measures the distances along its horizontal arms, as perpendicular to vertical as they can be on a roundish planet. But as the Earth's crust bulges upward, it also stretches horizontally. Think about drawing a picture of LIGO's arms on top of an inflated balloon and then squeezing the sides of the balloon. As the LIGO Observatory drawing rises and falls, it also stretches and shrinks. (Note that each piece of the drawing stays fixed on the surface of the balloon as it stretches.) That tidal stretching of the Earth's crust causes the 2-km distance between the midstations and the corner station of the observatory to change by tenths of a millimeter--a huge distance on the scale of LIGO's sensitivity to distance changes.
In fact, just like a sensitive seismic station that gets knocked off line by a huge local earthquake, LIGO's ultra-sensitive operation is knocked off line by the huge stretching of the interferometer--at least for now. In the figure at right above, we see a LIGO control signal applying forces to hold the mirrors at a constant distance from each other, even while the buildings move about underneath. This graph shows the control signal (in volts) plotted over a single day. These small control forces can counteract the effects of the tide for awhile, but eventually the small actuators--magnets glued to the mirrors driven by small currents in nearby coils--are overwhelmed by the magnitude of the tide. Then the control electronics "max out," the interferometer loses "lock" and the laser light stops resonating in the arms. The controllers then reacquire "lock" within a minute or so, and the laser light resonates in the interferometer arms once again. The control forces can then counteract the tides for another hour or two until they once more become overwhelmed.
This is no way to run an observatory! So there are special actuators in LIGO's mid- and end-stations that can move the entire seismic isolation system for each station by several tenths of a millimeter. There are also special actuators in the laser system that can make the laser wavelength stretch and shrink in unison with the Earth's crust. Computers will be used to estimate the future tidal distortions of the Earth's crust and to orchestrate these tidal actuators so that the actual control forces to the mirrors stay within an acceptable range. Once it is all tuned up, the interferometer should never lose lock due to Earth Tides!
The tidal estimation program was written during Summer 1999 by Eric Morgenson, then a rising Caltech sophomore interning at the Hanford Observatory. Unfortunately, it would take more than another year for the data to show up. The long data set from LIGO's second engineering run in November 2000, known affectionately as E2, gave five consecutive days of tidal cycling along the two perpendicular arms. This was just the data needed to debug the program! The plot at left shows the change in control force to the mirrors--averaged every fifteen minutes--in green. The estimated tidal drift of the 2-km arm length is shown in contrasting red. As is evident from the plot, the buildings are moving rapidly, at speeds up to 20 micrometers/hr (1 micrometer = 1/1000th of a millimeter), close to the raw tidal prediction. In fact, if we introduce a single "fudge" factor--multiplying the data by 0.8--the agreement would approach the error bars on the data. The remaining twenty-percent discrepancy will be tracked down eventually as more data is taken in future engineering runs. The most likely suspects are the precision of the internal calibration (estimated to be good to a few percentages), and some of the physical effects left out of the tidal estimation (like the effects of ocean tides and tidal tilts).
The next step in the story of the tides is to use the computed tides--"fudge" factor and all--and the tidal actuators to prevent lock from being broken by the tides, hopefully in time for testing during the E3 run this coming March.