1. VELO Testbeam 2006 Tracking and Triggering Jianchun (JC) Wang Syracuse University VELO Testbeam and Software Review 09/05/2005 List of tasks 1)L0 trigger using scintillation counters. 2)L1 trigger using number of hits and/or vertex. 3)Pattern recognition and track reconstruction.

2. 09/05/05Jianchun (JC) Wang2 Overview of Testbeam Events  The beam (p/  ) hits the detector sensors instead of (x=y=0) to align the telescope and monitor the stability.  Nearly all events in this scenario are with pass-through beam tracks. Only small portion of events have interaction. For alignment we prefer interaction events. The goal of trigger is to suppress the pass-through events.  The beam hit aluminum window at (x=y=0) to test the standard LHCb tracking algorithms. The trigger requirement in this scenario is similar. p/   Beam p/  SignalBackground

3. 09/05/05Jianchun (JC) Wang3 Beam Properties  Beam energy: 120 GeV p/  beam, 450 GeV p beam.  Beam time structure:  Spill: ~4.8 s beam per every 16.8 s.  Beam particles: ~10 7 particle/spill (~ 1 particle / 480 ns).  Interactions: ~2% ( ~1 interaction / 24  s).  Beam space profile is unknown. Following is just a guestimation.  Size: 20mm  20mm.  Shape: square distribution, flat in the center with slope at edges.  Beam direction dispersion: unknown.  The beam particle rate is low enough not to affect the trigger scheme. We need to know the beam space profile and direction dispersion to decide the scintillation counters.

4. 09/05/05Jianchun (JC) Wang4 Scintillation Trigger Scheme B1B1 A2A2 A1A1 B2B2 Trigger 1 = A  B Trigger 2 = (A 1  B 1 )  (A 2  B 2 )

5. 09/05/05Jianchun (JC) Wang5 Simulation On Scintillation Trigger  MC sample: 120 GeV  beam uniformly at x=(15, 35) mm, y=(  10,10) mm with no direction dispersion.  Two scintillation counters downstream at 2m from first station form a square of 200  200 mm 2 with a square hole at the center.  For schemes 1 and 2 the trigger efficiencies are ~94% and ~68% respectively if the width of square hole is 20 mm. (A 1  B 1 )  (A 2  B 2 ) A  BA  B

6. 09/05/05Jianchun (JC) Wang6 Summary on Scintillation Trigger  We have a preliminary configuration of scintillation counters. It is proved to work and optimized in MC simulation (JC + Tomas). However, we need real beam properties (space profile and direction spread) to decide the geometry.  The scintillation counters will be made at Syracuse (JC + Ray).  We have: enough 3” PMTs, HV power supply, CAMAC modules.  We have scintillation plastics, but may buy more if needed.  Light guide will be ordered.  The time budget of scintillation signal including amplifier, threshold, logic gates and cables is ~< 100 ns.  The system will be tested at Syracuse before shipping to CERN.

7. 09/05/05Jianchun (JC) Wang7 L1 Trigger Using Number of Hits  We can select interaction event at L1 (as a backup or supplement to L0).  The number of hits can be used as trigger, e.g. sum of maximum number of R hits and maximum number of  hits per station > 20. The efficiency can easily reaches 50% at noise level of about 0.1%.  The scheme needs to be optimized.  We will use large pulse height hits for trigger purpose. The efficiency can be much higher.  This scheme is very simple and does not depend on pre-alignment of the telescope.

8. 09/05/05Jianchun (JC) Wang8 L1 Vertex Trigger  Events where halo particle interacts in sensors can be eventually triggered by vertex searching algorithm [similar algorithm is used at H1 for vertex triggering].  Linear ‘ tracks ’ combining 2 hits on different planes are histogrammed + peak detection [a la non-interacting halo] R-sensors projection plane [R or φ sensor] ‘tracks’ [not all shown] Extracted from Tomas’ presentation at 07/22/05 VELO meeting

9. 09/05/05Jianchun (JC) Wang9 Summary On L1 Trigger  For L1 trigger we have two scheme candidates using number of hits or vertex finding. Both had been proved to work in MC simulation (JC+Tomas).  We will concentrate on number of hits trigger for now, as it is less dependent on telescope alignment. The method will be further optimized.  JC will implement the algorithm in Gaudi. The CPU time should not be a big issue as the algorithm is very straight forward. However, it will be measured.  It is projected to have a testable version by the end of October to the midst of November.

10. 09/05/05Jianchun (JC) Wang10 Pattern Recognition &Tracking  The official LHCb pattern recognition algorithm takes advantage of that the hits satisfy: r=p 1 z+p 2.  For testbeam track its hit r and z values are not linear anymore due to that it is not produced from the center (x=y=0). However the values follows:  Tomas developed an algorithm to reconstruct tracks. Instead of starting from most downstream station, this algorithm starts from the vertex. It uses fit of parabola function to find track seed and extend the track.  Tomas is optimizing the algorithm so as to use as much standard package as possible (thanks to help from Muriel).  JC and Tomas will implement the algorithm in Gaudi.