In preparation for the upcoming calorimeter calibration workshop, I wanted to organize a bit the history of past calorimeter calibration work. Here I summarize what I have found for the calorimeter tower gains calibrations starting a few years back (perhaps not back to the dawn of the efforts, but at least inclusive of the relevant years, I hope).

___________

Methods:

1. LEDs
   • Only in the barrel
2. Inverse slopes from ADC spectra
   • Plentiful statistics
   • Only relative gains within eta rings because spectra change with eta

   • Must use lowest portion of ADC spectrum because high end is dominated by showers from inactive material?

3. MIP peaks
   • Short lever arm
   • Reasonable statistics
   • Can provide absolute gains based on simulations / test beam data (SN 0433)
4. Electron E/p
   • Needs TPC; doesn't work at large eta
   • Reduced statistics (needs at least 1M tower-triggered pp events)
5. pi⁰ invariant mass
   • Complicated by dependencies of invariant mass (pt, eta?)
   • Reduced statistics
6. Isolated gammas (Tower vs. SMD energy)
   • Gammas identified by SMD

General notes:

• Barrel gain in the database has been defined in the calibrations to represent the coefficient to convert ADC counts into GeV. So the units are [GeV/ADC]
• Endcap gain in the database has been defined in the calibrations to represent the coefficient to convert GeV into ADC counts. So the units are [ADC/GeV]
• pp data provides the cleanest showers, so calibs are generally done in pp years

___________

Barrel tower history:

(much of it has been Calibrations)

Online calibs (changed to the HVs) have been documented by Stephen

• Run 3:
  • Marco did inverse slopes
  • Alex did MIPs tower-by-tower
  • Alex did E/p for entire detector (global)
• Run 4: (half EMC in place)
  • Mike & Adam did MIPs tower-by-tower, then ring-by-ring E/p fit to GEANT, then global E/p
  • Declared a 5% systematic uncertainty on absolute gains (see note at top of 2005 study)
• Run 5: (nearly whole EMC in place)
  • Repeat of Run4, some games played for ring-by-ring E/p
  • First with CuCu on west side only (starts 1/13/05)
  • Then with CuCu using new east side towers (starts 3/22/05)
  • Then with pp using full tower set (starts 2/1/05 and 3/22/05)
  • Declared a 4.5% systematic uncertainty on absolute gains
    
    • Suggestion to look more differentially (vs. eta and R = distance from tower center)
• Run 6:
  • HVs were modified in 2006 to give towers ~60GeV range (full scale of ADCs = 4096 - ~100 pedestal ~= 4000)
  • LEDs were used to calibrate online (set HVs)
  • Inverse slopes were used to calibrate online (set HVs)
  • Adam first did relative gains within eta rings by MIPs plus absolute gain from E/p ring-by-ring
  • It appears that Adam then did tower-by-tower MIPs and E/p, but never uploaded these to the DB
  • Adam reports from Run 6 that MIP peak position is known to about +/-5%.
  • Adam reports that gains seems to change systematically with time (stability, order 5%)
• Run 7:
  • Inverse slope outlier HVs adjusted online
  • Only changed outliers needed offline gains adjusted
• Run 8:
  • Not yet determined - in discussion
  • Will need to be wary of high eta tower gains
  • Examined:
    • MIPs (shows ADC value of MIPs too high at high eta = HV of towers too high)
    • Isolated gammas (shows BSMD clusters with too low energy at high eta = HV of towers too high, allowing low energy towers to pass trigger)
    • pi⁰s (shows invariant mass too high at high eta = HV of towers too high, causing E = C*ADC to be too high)

_________

Endcap tower history

• Full scale energy set to E_T (transverse energy) = 40 GeV
• Run 6:
  • Scott used MIPs (via showers with SMD hits)
• Run 7:
  • Scott used Inverse slopes to verify Run 6 MIP calib