===========================================
Minutes of Tracking Meeting, August 8, 2007
===========================================
taken by U. Husemann


Max Goncharov: Track Timing
===========================

- Regular COT t0 (written by Tom Phillips)
  o Measure t0 of single track w.r.t to (COT) clock
  o Measure timing difference between two electrons from Z:
    Perfect Gaussian with RMS 0.34 ns 
  o From W bosons: Perfect Gaussian with RMS 1.4 ns
- ctrefit vs. COT t0: 
  o ctrefit has 6 parameters, COT t0 has only 5
  o ctrefit 25% worse resolution, non-Gaussian tails
- COT t0 vs. phi: compare COT t0 and EM timing
  o COT t0: systematic effect in COT, 150% of nominal resolution
  o No effect seen in EM timing
  o Problem: track with bad timing may be thrown out of vertexing
- Wire-by-wire timing correction:
  o Done on jet20 data, one big run per period sufficient

- Aseet's concerns:
  o Cell-by-cell t0 fits are much more stable and should be good enough
    for correction
  o Wire-by-wire: need to check every single channel (30k)
  o Removing wires with bad fits reduced tracking efficiency 
    (more important for most people than good t0 resolution)
  o Current online correction: "wrong but stable"
    (mainly correcting for cable length),
    based on work by Ashutosh 5 years ago
  o Decision was made not to change online corrections
    New corrections are desirable offline
  o One run per period not sufficient for wire-by-wire calibration:
    oscillations, problems with single channels of TDC, ...

- Course of action:
  o Replacement of current offline t0 correction (text file) 
    with updated version
  o Antonio will run Gen7 production (Jet20, Z->ee) 
  o W mass group will provide W skim (time estimate: 2 weeks)
  o Slava will produce cell-by-cell t0 
  o Compare both files by running Production again
  o Check if wire-by-wire timing alignment within a cell is required
  o Metric for improvement: hit efficiency
 

All: Int10 Discussion
=====================

- Ulrich: still working on streamlining validation code
- Oliver: 10k test works, will invite people for a closer look (Friday?)
- Matt Herndon: double check Ben's new Silicon resolutions (fake rates)