BCAL Readout Segmentation Task group Meeting: May 19, 2011

Meeting Info

Time:

• 11:30 JLab
• 9:30 Regina

Location:

• CC F228

Vid-con:

• EVO
• JLab polycom IP: 129.57.66.12

Agenda

• BCAL Smearing (David L.)

Minutes

Participants

David Lawrence, Zisis Papandreou, Andrei Semenov, Irina Semenova

BCAL Smearing

We went over David's slides showing the status of the BCAL simulation. Much of the realism due to electronics, etc. is done in the mcsmear program. The core code was re-written (by Dave) and submitted to the repository. Several plots were shown indicating that the new code was able to (statistically) reproduce the results of the old code.

In order to aide in checking that the new code was working the same as the old, a new class called DBCALSiPMHit was introduced. This should be treated like truth information since no thresholds are applied to these hits and individual SiPM values will not be available in real data if summing is used.

The plots showing actual energy deposition (not just dark hits) were made using a 10k event data set meant to match the 2006 beam test. Namely, 90^o photons hitting the center of the bcal. The energy range was 160 MeV - 650 MeV. Differences with the 2006 data include:

1. Showers were evenly distributed in phi and so could overlap modules. For the beam test, the beam was centered on a single module
2. The energy spectrum in the simulation is flat while for the beam test, it would be bremsstrahlung (1/E)
3. The simulation modules are trapezoidal and the actual test module was rectangular

Note also that the "attenuated energy" means the energy would represent the actual energy if it were deposited right at the end of the module closest to the SiPM. The "attenuated energy" is therefore always a lower bound on the actual energy deposited.

Some discussion was had on the plots on slides 5 and 6:

• Andrei pointed out that the energy weighted timing is not accurate as the discriminator will fire when the leading edge gets above threshold.
  • Suggested recording energy in time chunks at ends of BCAL so a better estimate of the signal time can be made
• It was suggested that the ~17ns mean signal time be verified to be due to the time-of-flight plus signal propagation times
• Andrei suggested that the timing resolution may be significantly affected by the use of leading edge discriminators as opposed to constant fraction.

Discussion on slide 7:

• What is the source of the pedestal mean of 600 counts?
  • Is this a problem due to the current running through the SiPM package?
• Is the pedestal width due to the tile, or the pre-amp? If it is the latter, then it doesn't make sense to scale it by sqrt(16) for an array.

Slide 8 shows attenuated energy plus dark pulses for the new code after modifications to improve the realism were included. The large "humps" are due to the dark hits while the exponential tail is due to actual energy deposition. Thresholds were disabled for this plot so one could see the full spectrum that the discriminator would see.

Slide 9 presented numbers indicating that the 5% occupancy criteria to determine the threshold for the fine segmentation scheme would increase the event size by about 5.6% from the nominal 15kB. Discussion ensued:

• A recommendation was put forward to cap the total dark hit rate at 120 hits/event. This would keep us close to the 5% level
• Andrei suggested that the relatively higher occupancy seen by the inner layers as compared to the outer layers might motivate us to set a slightly higher threshold for the inner layers.
  • Some discussion was had on this point, but not fully understood
  • Andrei agreed to write up a suggestion for what criteria should be used to determine the threshold. Bandwidth will be a fixed parameter, but adjusting the relative values of the inner and outer threshold may be included
• Zisis pointed out that TDCs will only be used for the inner layer.
  • At the meeting, it was suggested this might save on bandwidth due to there being no TDC hits in the outer layers. In hindsight, however, one will want to record a time from fADC info for purposes of matching to the inner layers.

Slide 10 showed calculations of the attenuated energy distributions due purely to darkhits for both the fine and course segmentation schemes. The two curves indicate two different dark pulse rates. The higher one (41.1GHz) is what is currently in the repository. The lower one is from recent measurements by Yi and Carl.

• Numbers taken from Yi's talk should be verified that they represent the right operating temperature and gain
• Dave pointed out that the repository currently uses 3% for the cross talk and Yi advocates using at least 11%
  • Dave suggested changing this number in the repository as well, but didn't think it would have a large impact on these plots
  • Zisis agreed to changing the number in the repository provided it is a small effect. If not, then we would need to verify the temperature and gain settings for the data are what we expect to use for our operation mode

Other discussion:

• Dave mentioned that the KLOE code has now been modified to include a list of DBCALHit objects that were used to make the shower as "associated objects" in our reconstruction code.
• Irina suggested we have more discussion on efficiency so that we can at least come to a common understanding of how each of us defines it
  • Andrei requested that Dave send him the code used and he agreed to do so

Action Items

1. Investigate and possibly modify the simulation to include a time/energy profile of the shower in a cell in order to more accurately represent real data timing
2. Confirm 17ns mean time is understood for data sample shown
3. Understand source of 600 count pedestal and 8.5 count width in SiPM test data
   • Is the current leading to the 600 count create a problem?
4. Find status of BCAL discriminators (leading edge or constant fraction and is it a final decision)
5. Write down suggested procedure for determining thresholds based on fixed bandwidth requirement and circulate via e-mail
6. Modify BCAL calibration constants in repository to reflect most recent SiPM measurements
   • Confirm first that they correspond to expected operating mode
7. Send out details (including code) used to determine BCAL reconstruction efficiency