Difference between revisions of "SoLID Ecal Weekly 20230706"

Revision as of 22:13, 6 July 2023

Update from Carter

• Showed timing of SC-D minus SC-C, [insert slides here]. Fitted sigma is about 584ps. For LASPD Top - Bottom, about 2726 ps.
• Note that the bin size of the edge finding algorithm is 4ns/64 = 62.5 ps. With LASPD intrinsic timing resolution of 150 ps (based on previous study), we might see a couple of 100ps but not 1E3 ps level.
• Suggest reading through Jixie's work on LASPD timing from cosmic test, latest one reported on 11/16/2017 ECal meeting, look for correlation between timing and our GEM tracking position and see if we can do the same position correction
• Asked Mike to document (take picture of) all detectors, their readouts (picture of PMTs), then will collect data sheet of the PMT for their expected timing performance. Zhihong's preshower-tile is equipped with one of his PMT and he will send the info.

Update on PID

• Spencer working on ECal-only classical PID method, suggested increasing horizontal axis range to "capture" the full picture of the performance.
• Darren showed (on Monday) the momentum-binned classifier results. These however used the single classifier for all momentum bins. Since the classifier optimization could vary with momentum, suggested to develop an individual classifier for each momentum bin. After all momentum bins are studied and compared with classical results, then "lump" all events together to study ML with no momentum bin.
• Update on Thursday from Darren: SULI_Week5. Discussions below:
  • ECal: both CL and ML shows best performance for mid-p range, while the study is done with p < 5 GeV cut. This is puzzling, since at least for classical method the separation between electron and pion should be better for higher momenta. Will check this.
  • Cherenkov: e-efficiency seems to be low. Suggest doing the study with p< and p> electron Cherenkov threshold (typically at 17 MeV) and see if the low energy electron events failing to trigger Cherenkov is the reason. Note that Darren already have the study with p>1 cut which shows ~100% efficiency for Cherenkov. Perhaps can confirm by repeating the study for p< p_th (e) and p_th (e) < p < 1 GeV cuts, respectively.
• Next to do:
  • Study if all low energy particles can trigger the scintillators, and what are their energy deposit in the ShowerSum. If a correlation exists, it might help us to put a p_min cut since real data all need to pass through some trigger thresholds.
  • Mike and Ye continue developing data sample for "good electron" and "good pions".

Discussions on higher current runs

• lumi for SoLID is 1E37 -- 1E39, our beam test data with 70uA on LD2 target might get to 1/3 of the higher number here. So far we have been focusing on golden runs taken at 5 or 10 uA.
• data show nearly no effect from beam-on-target simulation. That is, the data spectra agree with e+pi simulation pretty well. Note that beam-on-target simulation is mostly (very) low energy particles and may not pass the scintillator or other detector trigger threshold cuts
• we need to study 70uA runs. Ye showed Shower L,T,R,sum for the beam current scan for LD2, but the scale of the spectra is not proportional to the beam current. The beam current scan for carbon target (lower thickness than LD2) shows better behavior Ye' slides Suggest comparing the two target data with the same current to further reveal the potential problem. Checking the FADC waveform might also help. (for example, if with 420mV ShowerSum threshold in TS4 we see carbon giving larger "apparent" peak integral threshold than LD2, one explanation could be that LD2 target pulses in the Shower blocks are narrower in time than carbon target pulses).
• note that Ye had reached 15% agreement between data and full sim for a 10uA LD2 target run. Would be good to check this for both LD2 and carbon and for all currents.

Update on GEM

• Xinzhan is back from FTBF and will focus on correcting GEM efficiency in the next 1-2 weeks.