1. First Data from the Linear Collider Alignment and Survey Project (LiCAS)
The ILC requires unprecedented accuracy, speed and cost efficiency for the survey and alignment of its more than 100km of beamline. Classical optical metrology in open air cannot meet these requirements. The Rapid Tunnel Reference Surveyor (RTRS), a self-propelled survey train, is intended to automatically survey a reference network in the ILC tunnels with an accuracy of 200 (500) microns vertical (horizontal) over each 600 m segment. A prototype RTRS has been built by the LiCAS collaboration. Calibration and first measurements are ongoing.
Armin Reichold, Patrick Brockill, Sigal Cohen, John Dale, Mike Dawson, Tony Handford, Mark Jones, Gregory Moss, Lee Antony Rainbow, Michael Tacon, Cecilia Uribe Estrada, David Urner, Roy Wastie, Stephanie Yang
The John Adams Institute for Accelerator Science, University of Oxford, Department of Physics, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, UK Johannes Prenting, Markus Schloesser: DESY Applied Geodesy Group, DESY, Notkestrasse 85, Hamburg, Germany
Grzegorz Grzelak: University of Warsaw, Department of Physics, Warsaw, Poland
The RTRS prototype consists of 3 measurement cars spaced 4.5 m centre to centre which travel on a rail along a tunnel wall. Each measurement car carries one measurement unit in its centre. Readout for each measurement unit is carried in its service car and a master car carries the common infrastructure. The RTRS has four measurement systems. An internal system of FSI lines (Frequency Scanning Interferometry, ~ 4m) and LSM lines (Laser Straightness Monitors) are both kept in vacuum to avoid air refraction. These allow the train to monitor its internal geometry. The third system is a set of external open air FSI lines (~ 40cm) which measure the 3D position of the reference markers for each car. The last system is a gravity referenced tilt sensor that measure each car’s rotation around the z axis.
LiCAS PURPOSE
LiCAS-RTRS
RTRS from LSM launch end
Parking clamp
Reference interferometer
Rotary vacuum seal
One of three drive motors
Internal FSI collimator and pellicle beam splitter
Unit 1 in its 6-D motion frame
Wall marker with retro reflector
View from top into motion frame
Wall wheel and parking foot
Vacuum equipment and LSM reflector end-cap
Master car
Master splitter
EDFA
FSI laser
Master computer
Service car
Stepper motor
Power supplies
Service car
FSI DAQ
Power supplies
DAQ computer
Splitter tree (behind panel)
DAQ Computer
LEAST SQUARES ANALYSIS: We follow D.E. Wells and Edward J. Krakiwsky, Hans Pelzer, Charles L. Lawson and Richard J. Hanson. In LiCAS, least squares analyses are used for calibration of the LSM, FSI and tilt sensors as well as for reconstruction to determine wall marker positions.
PERFORMANCE SIMULATIONS: The expected performance of an ILC survey was simulated using a two step process. At first a full optogeometric model of the entire measurement process for overlapping train stops was described in SIMULGEO and the predicted accuracy was fitted to a random walk model with angular correlations between steps. Due to the huge computational effort the Simulgeo simulation is only practical up to 20 train stops. Longer distances are simulated with a random walk model.
Right: random-walk Monte-Carlo Simulations over 600m and the average residuals against a straight line fit from many random walks. The deviation remains below 100 µm over 600m.
Left: SIMULGEO error propaga-tion for 20 stops with a fit of the random walk model and its extrapolation over 600m.
Left: Mean Fourier Spectra of random walk trajectories over adistance of 600m. Top is the spectrum of positions, bottom is the spectrum of residuals against a straight line fit
Right: Residua (top) and RMS (bottom) over 600m resulting from systematic errors caused by 5µm (sigma) calibration errors of all elements on a 4 car train with 25m car separation.
GLOBAL FORM OF THE DESIGN MATRIX IN RTRS: A full tunnel survey (~ 30 km) gives large matrices, just under 6000 blocks on the diagonal, not counting fixed parameters. Part of the total design matrix has a tridiagonal block form. Algorithms exist for quick inversions of these. We expect these to be faster and require less memory.
EXAMPLE CALIBRATION: A single car in the RTRS is moved and all measurement systems take data. The move is monitored at 4 points by a laser tracker, acting as an external witness.
SIMULATION RESULTS: All internal parameters seem to be obtainable in simulations. Data from the first two calibration experiments is being analysed now.
SIMULATION: From simula-tions, we know that we can determine the internal parame-ters from a least-squares pro-cess. This process seems best when at least 50 movements of the car are made, which leads to a design matrix of 2300 by 450 elements (only small part shown)
INITIAL: approximate loca-tions and orientations of com-ponents are taken from de-signs (orange).
FINAL: Best positions and orientations (orange) made to match external observa-tions. They agree with truth (gray).
(Non-zero elements appear as black blocks)
Laser Tracker
Simulations