1. DESY, Sep. 27, 2005 Warsaw University LiCAS Linear Collider Alignment & Survey A. Reichold, JAI @ Oxford for the LiCAS collaboration 1 Survey and Alignment of the ILC Alignment Errors

2. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 2 Overview The ILC survey and alignment process RTRS measurement principle for the reference survey Simulations of the survey process Reference survey SIMULGEO models for short distance (<100m) Random walk models for long distance Stake Out and alignment Interface to Placet Systematic errors in the reference survey Error Reduction How accurate can position alignment really be Cryo-module improvements fiducialisation build tolerances Survey and Alignment of non-Linac components Sources, DR, turnaround, BC, BDS, IP, spectrometer, polarimeter, dumps

3. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 3 The ILC survey and alignment process (in the tunnel) 1.Reference survey (the hard part):  stat <200  m/600m rapidly establish co-ordinates of regular array of reference markers along entire tunnel wall rapid = much faster then drift of tunnel rapid = as fast as possible since time = money hard = keep accuracy over large distance new instrument = RTRS (Rapid Tunnel Reference Surveyor) instrument developed by LiCAS group 2.Stake out:  <50  m any point Relate external accelerator component’s markers to reference markers measurement distance = across the tunnel diameter manually operated classical instruments could work but should be incorporated into RTRS if possible (speed, automation) 3.Alignment:  <100  m any point adjust position of accelerator element to get closer to nominal no automated process for cryo-modules exist yet manual action with a large wrench needed so far least accurate of steps so far

4. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 4 4.Fiducialisation (under development):  ≥300  m Relate external markers to relevant active centre line of accelerator element If necessary define an average active centre line if many elements determine the centre line as done in cryo-module Center line may dynamically change during cool down and transport. This is the biggest contribution to the fiducialisation error. 5.Build tolerances: Internal to an accelerator element there could be static variation of several active elements around the centre line (scatter of cavities in cryo-module)adjusted to  =100  m dynamic changes of elements with load, current, trim, external temperature, etc.  =???  m Externally there could be variations in target marker position when placing the markers repeatedly into their nests (dirt, dust, wear)  ≤1  m marker centring errors (markers are usually sphere mounted to allow for large acceptance angles by repointing. If target not correctly centred in sphere you get an error)  0.5  m The ILC survey and alignment process (outside the tunnel)

5. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 5 Notes about the process Notation: offset = one instance of statistical quantity called error error = sigma of the offset distribution Some offsets correlate within groups of accelerator elements all elements in a cryo-module share the same offset due to stake out and alignment  100% correlation External build tolerances also apply identically to all elements in a module but they are usually negligible The fraction of the fiducialisation error due to cool-down distortions is not the same for all elements but is systematically distributed to the elements  some correlation function dependent on module mechanics Offsets from reference survey are strongly correlated over very long distances. I.e. better then 200  m over 600m (see simulaions) All other offset sources are uncorrelated over distances > 1 cryo- module length

6. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 6 collider component RTRS concept Tunnel Wall Reconstructed tunnel shapes (relative co- ordinates) wall markersinternal FSI external FSISM beam LiCAS technology for automated stake-out process

7. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 7 LiCAS Measurement Principle External FSI System measures Wall marker location Internal FSI System  z. &  x,  y & ,  between cars Straightness Monitor  x,  y & ,  between cars

8. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 8 Build opto-geometric model of all measurements in a 6-car train and all reference wall markers using SIMULGEO Add up to 20 trains in advancing locations into the model  model consists of 20 trains measuring 26 wall markers.  total of O(10.000) elements and measurements with individual errors in the model Most wall markers get measured 6 times in overlapping measurements  this is how trains correlate with each other Perform error propagation: from: position errors of elements in the cars and measurement errors to : errors of wall markers, i.e. invert error matrix of rank N 2 = 10.000 2 Limit of this procedure is memory of computer  20 trains need close to 1 GByte and 34h on 2GHz CPU Reference Survey Simulations (short distance < 100m)

9. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 9 Simulgeo Model of RTRS only first and last car of 6-car train shown retro reflector Laser beam parallel to Gravity @ car 1 2 nd CCD faking clinometer 1 st CCD faking clinometer Laser beam parallel to Gravity @ car 6 internal FSI lines wall marker straightness monitor laser beam intermediate cars not shown LSM-CCDs for incoming beam LSM-CCDs for return beam

10. 10 Reference Survey Simulations (long distance >100m)

11. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 11

12. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 12

13. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 13 Stake out and Alignment Simulations Both processes so far only add a Gaussian random offset to the reference survey Width of these Gaussians is so far based on separate instrument errors and error propagation by surveyors. Assumptions:  horiz =0.3mgon,  vert =0.3mgon,  dist =0.1mm, DL between reference markers = 5m, Note: 1gon = 2p/100, 1mgon = 63  rad In future we will fully simulate an integrated stake out with SIMULGEO and determine the errors to see if we can decrease them.

14. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 14 Interface to Placet (licas_sim) Fortran program tested on Linux licas_sim produces the offsets of components given their z positions All internal tuning parameters (measurement resolutions) are in a control card file and can be modified by users Input: two flat text files (one for each arm of the machine) with one z position per line, representing the nominal z of the component to be aligned. Output: each line of input replaced with: z-original, x-offset, y-offset, z-offset All units are meters

15. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 15 Systematic errors in reference survey We have no measure of them yet Hard to measure because LiCAS “would” be the highest accuracy instrument Mainly to be determined via consistency checks during LiCAS test runs at DESY next year Potential sources of systematics exists in: miscalibration of components (predominant source) scale-mismatch between sub-systems drift errors on precision components systematic atmospheric effects Most of them would be addressed by calibration procedures Which systematic errors would be dangerous for LET?

16. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 16 Error Reduction Position Alignment: Today’s movers for cryo-modules manual coarse movers (lots of force needed) few cm range hardened steel feet on hardened steel plate If the estimated 100  m accuracy prove to be insufficient and money is available one can: Mount modules using bearings motorise adjusters achieve O(10 microns)

17. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 17 Error Reduction Cryo-module fiducialisation 300 microns is limit with current module design better is possible but at great cost part of 300 microns is currently uncertainty rather then error. I.e. expected scatter of beam line positions in a large number of modules.  if you measure each one of them internally (wire system) you can get better transport and handling is a large unknown factor

18. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 18 Error Reduction Cryo-Module build tolerances scatter of cavities in a string = 100  m < fiducialisation error  no need to improve offset of quads from beam line = 100  m This is currently not the leading error

19. DESY, Sep. 27, 2005 A. Reichold, JAI @ Oxford for the LiCAS collaboration 19 Survey and Alignment of non-Linac components We know neither specs nor methods yet We are working on a document to collect these Would be very useful to prioritise the list by knowing which part is most sensitive Input from WG1 of highest importance