D0 RUN2B TRIGGER MEETING
		       ========================

		   8:00-9:00am (CT) 15 March, 2002
			       Far Side


Rich Partridge		Quantization Study for L1 Cal Trigger
--------------
o Question: for a given amount of noise, how big should the lsb be?
o Noise Model
  - Gaussian distrib
  - uncorrelated b/w cells
  - Jet energy in 36 cells
  - Energy Distrib models
    a) constant over cells
    b) exponential
o Quantization Error
  - E-Dist a) quantization effects begin to be sizable when lsb =
    3-4*noise per cell
  - E-Dist b) less pronounced biases - but still setting in at
    3-4*noise 
  - Bias in E-measurement --> offsets of up to 2*intrinsic noise
o E=0 bin not centered (offset by 1/2 a bin)
  - Large biases possible
  - this is probably the most important effect
  - need to account for small negative energies
    . add a small pedestal?
o Conclusion
  - using lsb = sigma(noise) should be a good choice
  - need to find out what this is
    . from detector sums (do we have this as an rms)
    . from digital filter (will this be important?)



John Parsons		Summary of Recent L1Cal Hardware Discussions
------------
o Signal Duplication for TABs
  - strong desire to avoid using more than two copies
  - Consensus: duplicate at ADF level --> two output cables per board
o ADF to TAB Data Transmission
  - Consensus: use commercial Channel Link chipset for this
    . note: no increase in latency b/c of this
o Which Channel Link
  - Consensus: 48:8 chipset preferred b/c of data rates to chipset and
    on cable
o Cables
  - investigating cable possibilities: vendors contacted
  - important questions: do we need halogen-free cables
    . talk with Rick Hance
o Data Synchronization Issues
  - these will be discussed in an upcoming meeting


Jiri/Emmanuelle		    Status of L1Cal Simulation
---------------
o Documentation
o Inputs
  - TT from customized tsim_l1cal
    . allows different rounding/truncation schemes to be studied
    . rcp setable
o Triggers Implemented
  - global, electrons, jets
  - ICR can be included or not
  - each trigger definable by rcp
o Output
  - single Ntuple
o RoI Definitions
  - rcp setable
    . overlap between RoIs
    . jet energy summing region
    . electron isolation definitions
    . can choose Atlas algorithm for electrons
o First Results
  - QCD events
  - No. of jets vs. declustering region: 3x3 vs 5x5
    . very small difference b/w the two
  - ICR: compare Pt-trigger vs Pt-reco
    . no difference b/w using or not the ICR in the jet def's
    . could this be a weighting problem in d0sim
  - Atlas algorithm: electron maxima defined by 1x2 or 2x1
    . standard 2x2 algorithm seems to be preferred


Meenakshi Narain	L1CTT FPGAs
----------------
o New simulations show: Increase in No. of equations over 2a x10-x30
  . this is more than was assumed in the CDR
  . FPGAs that could deal with this (>2M gates) cost in the range
    $2K-$5K 
o This will change cost estimates by a substantial amount
  . need to understand how to fold this into the upcoming reviews