BCAL Readout Segmentation Task group Meeting: Jul 14, 2011

From GlueXWiki
Jump to: navigation, search

Meeting Info


  • 4:30 JLab
  • 2:30 Regina


  • CC F326-327

Telephone: (JLab cyber-event prevents vid-con)

  • U.S. and Canada: 866-740-1260
  • Access code: 3421244




Andrei S., Zisis P., Irina S., David L., Elton S., Matt S.

Review status and form timeline for decision on segmentation

No new material was available to review. A combination of various things including vacations, the cyber-event at JLab and work accumulating from other projects limited time spent on this task over the last 2 weeks. The purpose of the meeting was to review the current status of the task group's mission and to form a definite plan on how to proceed. To help guide this discussion, David put together 2 slides reviewing the work that has been done on this topic since March and summarizing what has been learned (from his viewpoint). Here are some discussion points:

  • Prior to starting, Andrei noted his objection to many statements and conclusions in David's slides
  • The summary slides generated significant discussion:
    • Dave suggested that evidence gathered up to now points to (albeit inconclusively) that the polar angle θ is the only reconstructed parameter that might be significantly affected by the choice of segmentation.
    • Andrei opposed this statement with the strong argument that HDGEANT/reconstruction code (that came from Indiana U. and was tested by Dan Bennett and later by Irina and Andrei with better statistics and more careful fitting procedure) contains incorrect simulation of the time signals from BCal as well incorrect accounting for the BCal energy resolution because of the calorimeter sampling fraction. Without correct handling of the timing information AND proper energy smearing in the mcsmear code, none of these "code testing" results so far obtained must be considered a reliable comparison of the two segmentation schemes (even though all of them are favorable to the fine segmentation scheme); Andrei made this "legal statement" very clear before in every presentation of "code testing" results, and this statement must not be omitted in the future references.
    • At the June 17th meeting, Andrei and Irina proposed a method (including the draft of the code lines) to account correctly for timing and energy deposition information (see minutes); as one of the principal creators of the simulation/reconstruction code, David was assigned to implement promptly the suggested method in to the code, and Elton was assigned to review Andrei's proposed procedure for calculating sampling fraction functions.
    • At the June 30th meeting, David demonstrated a big progress on implementing of Andrei/Irina's timing scheme into the code, and it was stated that the code might be finished and the results might be available by mid to late July at the latest (see minutes).
    • Unfortunately, the holidays, cyber event, too thorough refinement of the exact shape of SiPM pulse and LE discriminator thresholds, and no decision made on procedure for correct sampling fraction simulation delayed the finishing of the ready-for-the-test HDGEANT/reconstruction code. Due to the declared time constraint, David proposed to make a decision to go with the course grained plan unless new evidence to the contrary was obtained in the ongoing timing studies by the meeting scheduled for July 21st.
    • Andrei counter-proposed that we concentrate on the fixing of the major part of the simulation code (without scattering of our resources on non-vital refinements) and make the fine/coarse-segmentation decision on the basis of correct (though not-refined) results from the simulation that might be available as soon as 1-2 months (if we work hard). To achieve this, Andrei suggested that UofR group start a generation of lookup table for the correct sampling fraction smearing without any further delays.
    • Zisis made a compromise proposal that we make the decision to add one more ring pending strong evidence emerging over the next week. This was based on discussions with experts at Regina as well as with a KLOE expert. This generated considerable more discussion:

Here is David's attempt at an objective summary of the debate acknowledging that it may contain David's own biases:

Side "A" believes the finer grained segmentation will record more information and is therefore the safer solution in the absence of firm guidance from simulation

Side "B" believes the finer grained segmentation will lead to a smaller signal to noise ratio on average and since the threshold is set by the noise level, more energy will be thrown away by threshold cuts potentially leading to a loss of information.

This issue is complicated slightly more by the fact that Eugene left open the door for adding another ring to the inner BCAL region by stating it could be done if "compelling evidence" indicated it was needed. No further guidance has been given yet leaving how high the bar of "compelling evidence" is to some interpretation.

    • Matt asked if the electronics could be designed to allow the summing to be determined by jumper settings. This could allow the board design and the simulation studies to go forward in parallel.
    • Elton said that had been considered, but was deemed to add significant enough complication as to make it highly undesirable. He did, however, suggest that perhaps a 2-option solution with single jumper to select which option might be less complicated.
    • The question of longitudinal (radial) depth in segments in the KLOE detector. Zisis recalled it was 5 deep, but wanted to verify the number (it was later circulated via e-mail that this number was correct via. NIM A482 (2002) 364)
    • Some discussion developed on the sensitivity of the polar angle to the sampling fluctuations. Matt recalled from simulation studies he and Blake did that this was a mild effect. He volunteered to track down the references and circulate them.

We came back to the discussion of schedule. It was reiterated that this subproject is 6 months behind and a delay in the segmentation decision delays the board design which further delays this part of the project schedule. It was noted that current studies of an increased rise time are underway and if they show a problem going from a 10ns to a 15ns rise time, the board design could be delayed for other reasons giving us more time for simulation studies.

It was suggested and agreed to that if we needed 1 more month to complete the current simulation studies that we should take that month to come for a more solid basis for a decision. The discussion was left there to be continued at the next meeting.

Action Items

  • Verify KLOE used 5 rings total (Zisis)
  • Circulate info on polar angle sensitivity to sampling fluctuations (Matt)
  • Implement timewalk corrections and continue timing resolution study (David)
  • Present results on extra showers from course segmentation (Andrei)