The CLEO CsI Calorimeter

1
The CLEO CsI Calorimeter

• 15 Years of Physics with Photons

2
hc (1P1 charmonium) Discovery
?(2S)??0 hc??0(??c)
5.5? significance
M(hc)3524.4?0.6?0.4 MeV (consistent w/spin-wtd
?cJ avg)
3
Requirements

• Early 80s detector landscape
• Good tracking poor calorimetry
• Mark III, CLEO I, Argus
• Poor tracking good calorimetry
• Crystal Ball, CUSB
• CLEO II have both!
• EMCal must fit outside 1m radius tracking
  chamber
• Magnet coil outside EMCal, so EMCal must be
  compact, work inside B-field

4
CsI(Tl) Properties
Light Output
?E/E ()
5
Hermetic Coverage
6
CLEO III/c Layout
48 rings, 128 per ring
RICH
820/endcap
Drift Chamber
7800 crystals overall
7
Calibrations

• Electronic channel level
• Pedestals Gains on each of 3 ranges
• Crystal level conversion factor
• Shower level Energy Scale

8
Bhabha Calibration

• Assume linearity of light ? energy
• Obtain a single constant per crystal gives
  relative gain
• Select ee- evts each e? has Ebeam
• Minimize shower energy resolution
• 7800?7800 sparse matrix equation (non-zero near
  diagonal)
• Once determined, why should they change?

9
Light Collection Degradation
Crystals migrate from the top band to the bottom
band, then stabilize.
Cause lucite-CsI glue joint opens up.
10
Absolute Energy Scale

• ?0???
• ????
• ??-?
• ?(2S)???cJ

Fit huge combinatoric background subtract
E?
11
Current CLEO SCal
12
Performance
Energy resolution
Angular resolution
M?? (GeV)
13
Non-photon rejection

• Energy deposited near tracks easy to reject on
  that basis
• Nuclear intns cause splitoffs separated from
  central shower matched to track
• Splitoffs cause bgd for ?0, ????
• Splitoffs make neutrino reconstruction (missing
  energy) more challenging
• Several algorithms developed for rejection
• Based on lateral shower profile- collimated or
  not?
• Proximity to track entry points
• Energy

14
Lepton ID Example ?(2S)???- J/?, J/??ll-

• e? deposits all energy
• ?? deposits 220 MeV (min. i.) with Landau tail
• Form E/p variable shower energy/momentum
• Peak near 1 for e?
• Only small tail from ?? at high E/p
• ?? peak near small E/p

MC
Data
15
Example ?(2S)??0?0 J/?, J/??ll-
Very clean, well-modeled
???0?0?0
16
Conclusions

• W/careful design, a CsI(Tl) EMCal offers
  excellent resolution in energy (1.5-8) angle
  (3-15mr) for E0.05-5 GeV
• Long term stability demonstrated (glue joints!)
• Preserve energy resolution w/careful summing
  angular resolution w/MC corrections to c.o.g.
• Scale calibration a challenge, but can achieve
  
• Shower shape, track-shower matching both useful
  for isolation splitoff rejection
• Payoff in PHYSICS ?-lines, ?0 ?, e? ??,
  ?-reconstruction

17
Crystal Testing
18
Readout

• 4 diodes/crystal
• Local preamps, 4 crystals/card
• Externally summed
• Shaped
• Digitized (3 scales to preserve dynamic range)
• Sparsified
• To tape

• Disabled diode compensation
• CLEO has 4 diodes/crystal, can disable
  compensate via downloadable settings to crates
• Output gets amplified up accordingly
• After 15 yrs, CLEO has 251/7784 crystals w/?1
  diodes off because they died or became noisy
• 240 (1 disabled), 11 (2 disabled), 0 (3 or 4
  disabled)

19
Dead Diodes

• Disabled diode compensation
• CLEO has 4 diodes/crystal, can disable
  compensate via downloadable settings to crates
• Output gets amplified up accordingly
• After 15 yrs, CLEO has 251/7784 crystals w/?1
  diodes off because they died or became noisy
• 240 (1 disabled), 11 (2 disabled), 0 (3 or 4
  disabled)

20
How many to sum?
Optimize additional energy vs additional noise.
21
Position/angle determination

• Correction to centroid
• 0 at ctr
• 0 at boundaries
• not symmetric due to 50-mrad tilt away from
  vertex-pointing staggering of front faces
• smaller effect at larger energy because shower
  spreads more, which helps centroid accuracy
• peak correction typically 5-10 mm

22
Shower Energy Scale

• Which do you want?
• ? energy peaks at right place
• ?0 mass peaks at right place
• You cant have both, because
• Line shape has low-side tail (leakage!)
• W/correct ? energy, ?? mass peaks LOWER than
  M(?0)
• CLEO chose ? energy peak to be accurate, allowing
  ?0 constrained fit to fix the bias