We have known for a few years that the low-frequency ground vibration at the LIGO Livingston Observatory is significantly greater than that experienced at our Hanford site. As shown by the figure at left, in the 1 to 3 Hz band, the ground typically shakes about seven times more at Livingston than it does at Hanford. The graph is a cumulative histogram of the fraction of one-minute duration measurements of r.m.s. velocity in the 1 to 3 Hz band yielding a result greater than the value on the horizontal axis. Livingston's noise is also worse than Hanford's at even lower frequencies, all the way down to 0.1 Hz.
This extra motion makes the Livingston detector's mirror-positioning controls work much harder to acquire and maintain optical resonance in our four-kilometer Fabry-Perot arm cavities. Consequently, the bulk of the Livingston data taken during the first three LIGO science runs came from night and weekend operation, when forestry and other local human activities were not in progress. Our detector only worked reliably when ground noise in the troublesome 0.1 - 3 Hz band was below average, comparable to the levels at Hanford.
The new Hydraulic External Pre-Isolation (HEPI) system is a replacement for fine
and coarse actuation stages that support the LIGO payload in each tank, and was
designed to augment the existing seismic isolation system.
HEPI had been under development at Stanford for use in the proposed Advanced LIGO
detector upgrade, but about two years ago we realized that we had to accelerate
our work and install an external stage in Livingston to allow round-the-clock
During the first two weeks of August 2004, we tested HEPI performance by
using the single-arm interferometer configuration of the LIGO detector.
The graph below shows a number of interesting outcomes.
The green solid trace is the velocity spectral density of the arm length changes due to vibration that are being corrected by feeding back directly to the test masses, and so represents the effect of troublesome ground vibration, with HEPI turned off. The dashed green line is the accumulated root-mean-squared velocity, (in units of meter per second) calculated right-to-left. So, nearly 4 µm/s of disturbance is present, and other statistics indicate that the day these data were taken was at the 95th percentile in ground noise. We know from experience that our detector won't work in its two-arm gravitational-wave-hunt mode on days that are this noisy. The 1 µm/s dashed magenta line shows the RMS level above which locking is difficult.
The blue lines show the same things, with HEPI turned on in the end and
inner test mass payloads in our X arm. Noise is reduced along the entire
troublesome band, and the RMS velocity is brought down to a level that would
make operations possible!
It has taken two years' worth of the minds and labor of dozens of LIGO Lab and LSC collaboration members to bring HEPI from the prototype stage to where we are now, a fully installed, almost completely commissioned system at Livingston. Here is a brief summary:
|Onwards!||to the LIGO Chronicles at Caltech||Or Return!||to the Newsletter Front Page|