1. 2005 LAT-COLLABORATION MEETING Gamma-ray Large Area Space Telescope A walk through the LAT Tracker towers alignment Nicola Omodei INFN-Pisa

2. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 cucina italiana I – tower ingredients 19 Al honeycomb 38 C-fiber facesheets 38 C-C structural closeouts 38 C-C MCM closeouts 36 Multi-Chip Modules 576 SSD  55K channels  228  m pitch 192 3% X0 W tiles 64 12% X0 W tiles 36 kapton bias circuits 4 C-fiber sidewalls ~ 1000 screws (several types) 8 kapton flex cables glue, paint, tape ……

3. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 cucina italiana II the course and the flavour

4. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Silicon Strip Detectors – japanese devotion SSD reference crosses Alignment of strips wrt wafer cut: displacement of each cross from nominal distance is measured for both axis under the INFN microscope/CMM Any displacement has an RMS ~ 2.5  m Requirement of maximum displacement < 20  m comfortably met by each sensor 8.95 cm

5. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Ladders assembly and metrology The excellent SSD alignment allow a simple and cost-effective approach to ladder construction: –silicon wafers are wet with glue on the edge and positioned in the assembly jig –each wafer is pushed against the adjacent wafer and against a reference wall machined to high precision –once in place the wafers are held in position with vacuum and cured at RT No need for lengthy and expensive CMM alignment 8 reference crosses position are measured and their displacement from the best-fit line is plotted All RMS are < 2  m and the maximum displacement is typically ~ 5  m

6. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Trays assembly strategy Trays are made by gluing composite materials that cannot be machined to extremely high precision The reference pins for aligning the tray to the assembly jig are redundant –the position of each pin is measured –for each tray the best 3 pins are chosen to define the tray position in the assembly jig reference frame –Ladders are referred to their assembly jig and handled with a tool that is in turn referred to the tray assembly jig tray Ladder assembly jig tray assembly jig (reference frame) Ladder handling tool Positioning and gluing of ladders on tray

7. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Flight tray metrology The ladders displacement (X/Y/Z) on trays has an RMS ~ 20  m SSD inside ladders are controlled to a much better precision The silicon layer can be considered as a single, monolithic plane and its position in the tower can be controlled offline, at 1 st order, with 3 parameters (  x/y, , 

8. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Tower assembly and metrology Trays are slit in position with a crane and stacked into the assembly jig (top to bottom and alternatively rotated at 90 o ) Each tray position is referred to the assembly jig (absolute reference frame)  displacements in XY or Z are averaged out and do not sum The complete tower is measured under the INFN CMM Grid simulator CMM head

9. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Tower metrology Typical tower stay-clear verification All towers are well within specs Nominal position Stay-clear (375  m from nominal in all directions)

10. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Tower metrology with CR Cosmic Rays can be used to study the trays alignment –Vertical shift, horizontal shift and rotation around z. –Cuts on events can be used to study the alignment of the single ladders: the results is that the silicon plane (4 ladders) is a monolithic structure (within 20µm): at this stage disalignments are only shifts and rotation of a tray wrt the others! Iterative procedure: –Starting from a MC ideal geometry –One iteration. For each event: Fit a reference track for each view (2 fits) Cut on the chi square to reduce the cpu time (35% events selected) For each plane, refit the track with clusters removed on both planes of the tray (36 fits)

11. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Residuals vs. slope Aligns: horizontal (  to strips) vertical X4 X3 X2 X1 X0 real position ideal position res =  x +  z · cot(θ) θ horizontal displacement: 157  m vertical displacement: 81  m Horizontal and vertical alignment

12. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 X Residuals vs. position in other view Y, ideal position Y, real position rotation around z: 0.54mrad A straight line fits well all 4 SSDs in all 4 ladders  all 4 ladders are well aligned  tracks do not see independent sensors

13. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Iterative alignment procedure MC geometryAligned geometry 25 k50 k75 k100 k ~ 2 hrs data taking Minimizing the residuals: iteration procedure, till Either the new is identical to the old geometry Or the new geometry is identical to any in the sequence of geometries generated during the iteration (there can be periodic sequences of n geometries). The precision reached depends on the number of events analyzed

14. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Coordinate shifts X The tower jigs aligns the pins that have been used to align the ladders (the best reference) When the pins on the alignment wall of the tower jig are removed, the sidewalls partially realign the trays to their external dimensions with a max alignment loss ~0.1mm Sidewalls assy Vibe test TV test Sidewall removal Shipment to SLAC

15. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Comparison between Pisa and SLAC runs No relevant differences before and after the shipment 20 µm

16. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Back to the ideal world of MonteCarlo Residual from real data after offline corrections: rms = 137  m Residual distribution from MC data : rms = 124  m all effects from ladder, tray, tower integration are minimized through careful assembly such misalignments are controlled in the offline with few, small parameters: 1728 constants for all 576 layers with 3 parameters each, rather than a maximum of 6 parameters for all 9216 SSD (>55K constants!) what remains after the correction is very similar to MC data, where only intrinsic tracking fluctuations hold (single hit resolution and multiple scattering)

17. Nicola Omodei @ INFN PisaSLAC-August 30, 2005 Conclusion Alignment and geometry extensively measured for all assembly and test phases Many different components are integrated into the tracker: –the excellent intrinsic precision of base components (SSD) is fully exploited and maintained in higher level components (e.g. ladders) –assembly procedures overcome limitations from specific materials (composites, glue) and deliver integrated units (trays, towers) with monolithic performance (this is also valid for electrical properties) The design, the precision of components and assembly is such that the relative small misalignments and rotations of layers in a tower can be controlled by simple offline corrections based on few parameters The same performance homogeneity hold among all tracker towers and is crucial for a smooth integration with the other subsystems and a good calibration of the LAT, with positive impact on the science performance This is the same difference between 200ml of olive oil, 1 egg, 1 spoonful of lemon, salt and a good, creamy mayonnaise!