DIRC Prompt Reconstruction

DIRC detailed Monitoring

This document evolves from the comprehensive initial note by L. DelBuono, L. Roos, M. Zito prepared in the context of the DIRC OEP. Here we sort out the quantities which will be best monitored within the Prompt Reconstruction. This is the proposed scope for the so-called detailed monitoring for the DIRC (see D. N. Brown's HN for the vocabulary).
Recall the DIRC characteristic numbers: Nch=10 752 phototubes (PMT) are fed in groups of 8 into Nanalog=1 344 analog chips; two of those send signals to Ntdc=672 digital chips. The chips are mounted on Ndfb=168 Dirc Frontend Boards (DFB) which service 64 PMTs. Each of the 12 DIRC sector houses 1 frontend crate with a DIRC Crate Controller (DCC), so Ndcc=12. The PMTs are supplied high voltage in groups of 16, thus there are Nhv=672 high voltage lines. There ar=672 high voltage lines. There are Npm=896 PMTs in one sector. There is 1 bar box housing 12 quartz bars in front of each sector for a total of Nbar=144 bars. A typical event is expected to give Nhit=300 hits.

The aim is to provide the experimenter on shift with a picture of the DIRC response to physics events. We broadly categorize the quantities of interest into event related quantities, quantities which tell something about the optical transport of the photons and those more related to the detection (mainly electronics and phototubes) and data acquisition processes. Quantities which need to be followed over time are listed at the end.

The processed events will stem for various trigger types. We don't know yet their number Ntrig, but assume it is <=10. Various categories of events or event types will be useful to determine specific performances: Bhabhas (radiative or not), muon pairs, hadronic events, etc. We assume there will be 5 or less event types.

Available to the user, will be histograms, tables and scalars. Some will be available by default while others, only on demand when further details are needed.

In what follows a run is a block of contiguous data in time. It is anticipated these blocks will last about 30 minutes. Most plots are expected to be refrests are expected to be refreshed for each run/block. Exceptions are plots which look at small sections of the DIRC (plots made for each bar of fraction of a bar, plots made for each tube) and profile plots which follow measurements with time (typically one entry/run). The profile plots are assumed to contain 1000 bins (1 week including false starts).

Some caveats:

• Associations of DIRC hits with Calorimeter clusters may not exist.

• Events To be checked: counting rates per event, geometrical positions of the hits, timing. Do checks overall. Repeat some checks for hits associated to tracks, hits associated to crystals (and not to tracks) and completely unassociated hits. Some checks for various kinds of triggers/events.
  • Counting
    
    • N(tdc_hits)
      
      • all hits: 1 * (1d, 100 bins, bw=5)
      • unassociated hits: 1 * (1d, 100 bins, bw=5)
      • Tables: mean, sigma for all, /trigger type, /event type. For instance in sectors w/o track in Bhabha or mu pair events.
      • sin(dip) associated hits in Bhabha events: 1 * (1d, 10 bins, 0.2)
    • N(tdc-hits)/sector
      
      • 12 * (1d, 100 bins, bw=1)
      • Tables: mean, sigma for all and some trigger/event types. For instance in sectors w/o track in Bhabhr instance in sectors w/o track in Bhabha or mu pair events.
    • N(adc_hits), all : 1 * (1d, 100bins, bw=5)
  • Geometry and frontend sections
    
    • tdc_hit maps : scatter plot 2d, >10752 bins (for intuitive display). Zoom on individual sector possible.
      • 1 for all hits
      • 1 for hits associated to tracks
      • 1 for unassociated hits
      • 1 for hits associated to crystals (and not to tracks ?) MAYBE
    • From 1st plot (global hit map), extract hit maps by gathering bins on demand for
      • sectors 1 * (1d, 12 bins)
      • DFBs 1 * (1d, 168 )
      • HV lines 1 * (1d, 672 )
      • Digital chip 1 * (1d, 672 )
      • Analog chip 1 * (1d, 1344 )
      • DFB zooms <=14 * (1d, 64 )
  • Timing
    Unless specified otherwise, time means T(L1)-T(tdc). BCO means time of bunch crossing at which event occurred. T(L1) jitters around BCO within trigger resolution.
    • Coincidence window
      • global : 1 * (1d, 150 bins, 10ns)
        associated hits: 1 * (1d, 150 bins, 10ns)
        trigger resolution, selective R/O performance
      • on demand /sector: 12 * (1d, 150 bins, 10ns) timing differences between sectors
      • on demand /trigger type: Ntrig * (1d, 150 bins, 10ns) resolution of given trigger t10ns) resolution of given trigger type (MAYBE)
    • Event duration:
      We strive to estimate a (reliable) reference time, T_ref for each event,
      plot
      time - T_ref (all) : 1 * (1d, 40 bins, 5ns wide),
      time - T_ref (associated) : 1 * (1d, 40 bins, 5ns wide),
      Table
      acceptance time window for Cerenkov photons, accidentals.
    • BCO from DIRC.
      Plot: implementation to be defined
      Table: BCO quality
    • Global t_0 calibration performance. Assuming other detectors (or DIRC) give BCO,
      Plot: associated hits: (t_meas-t_expected]
      1 * (1d, 50 bins, 0.5ns)
      Tables:
      list of PMTs outside range
      
  • Charge. Cumulated ADC spectra in relative units
    • plots:
      all hits : 1 * (1d, 100 bins, )
      per sector: 12 * (1d, 100 bins, )
    • Tables: PMTs with abnormal working points
  • PID. For associated hits
    • Theta_c vs P : (2d, nx=50, ny=50)
    • Theta_c_meas - Theta_c_exp for Bhabhas
      
      • Plots 10 (dip angles) * (1d, 50bins, )
        "Phi_c" acceptance (to be defined)
        
      • Tables: mean, sigma of associated hits vs dip
    • Similar for mu pairs (at least initially)
    • Images from unassociated PMTs.
      Count, tally. Don't kciated PMTs.
      Count, tally. Don't know how to proceed yet. Statistics on
      • images reconstructed w/o using tracks,
      • images recognized among unassociated hits.
  • Background
    Use above statistics on unassociated hits (i.e. already mentioned).
    Quartz scintillation. To be defined. Guess:
    theta - theta_c vs phi: 2[a,b] * 2[c,d] * (2d, 20 * 20bins), for hits close to a track (a) with measured theta_c, (b) with no theta_c measured (theta_c=0), (c) either on time or (d) not.
    Same with less bins each bar: 144 * 2 [a,b] * 2[c,d] (2d, 20 * 5 bins).
• Optics
  • Associated hits, n_tdc/bar : mean, sigma
  • Path length in quartz for attenuation and, from that reflectivity of mirrors, glue joint effects, wedge. Invent optimal method,
    1 global + 12 +
    144 * (1d , 50bins, )
  • Nb-associated hits vs theta for Bhabhas and mu pairs 144 * (1d , 50bins, ) (could be used instead of the previous one)
  • Water effect assuming no change in quartz
  • Light catcher effects (MAYBE LATER)
• Detection In factory mode, all that should be taken care of in OEP (fast monitoring) and above.
• Data acquisition
  • Error codes if available (tbd)
  • Buffer ove Error codes if available (tbd)
  • Buffer overflows (tbd)
• Follow with time All plotted quantities vs time (i.e. reference time of block/run) are drawn with appropriate error bar.
  • 1 N(tdc_hits)/ event
  • 12 N(tdc_hits)/sector/event
  • 1 Fraction of tdc hits non associated to tracks
    
  • 2 Nb of unusable PMTs (a) too hot; (b) dead; (c) faulty selective readout
  • 1 Event duration
  • 1 Offset between LED and beam timing
  • 12 Same per sector
  • 1 [t_meas - t_propag]_average
  • 1 BCO_dirc - BCO_other
    
  • 1 average PMT gain
  • 1 average PMT noise
  • 1 average PMT efficiency
    
  • 12 scaler counting rates
    
  • 1 Cerenkov angle (Bhabhas) [spread is scatter among sectors]
  • 1 Resolution of Cerenkov angle (Bhabhas)
  • 1 Cer. angle (mu pairs)
  • 1 Resolution on Cer. angle (mu pairs)
    
  • 1 Average attenuation
    
  • 1 representative efficiency and purity (same plot) for pions from Ks
  • 1 same for pions from Lambdas
  • 1 same for protons from Lambdas