The following is a note which tries to describe how Trigger Tower energies are received and understood by the various systems in Run I and Run II. It is NOT an exhaustive summary of all types of data handled by these systems. It DOES contain some information that we might NOT want to share with the general public yet (specifically in the Run II L1 Cal Trig section). For the Run I (and Run II) L1 Calorimeter Trigger ------------------------------------------------- See D0 Note 706, on the web at: http://www.pa.msu.edu:80/hep/d0/ftp/run1/l1/caltrig/d0_note_706.txt and D0 Note 1680, on the web at: http://www.pa.msu.edu:80/hep/d0/ftp/run1/l1/caltrig/d0_note_1680.txt Input data for each Trigger Tower is the analog trigger pick-off of EM and Hadronic energies (from the BLS cards), which are scaled to Et before digitization on the L1 Trigger Framework CTFE cards. These are independently digitized into separate 8-bit values, each with a 0.25 GeV per count energy scale and a FIXED pedestal value of 8. These 0.25 GeV bins are centered ("rounded") at the nominal value (i.e. Et between 0.125 GeV and 0.375 GeV is digitized as 0.25 GeV). We call these the EM and Had "ADC Counts". For each Trigger Tower, the EM and Had ADC Counts are fed independently to lookup memories, which produce new 8-bit outputs. We call these outputs the EM Et and Had Et for each Trigger Tower. These lookup memories do the following: - apply a symmetric low-energy cut (max of 0.5 GeV or 2.5 sigmas of electronics noise) - shift the fixed 8-count pedestal to a new value which is a function of eta (for purely technical reasons) - apply z-vertex correction (not used in Run I) - they DO NOT modify the energy scale The output of the EM Et lookup memory by itself is used in two ways: - it is driven to the EM Et Adder Tree to produce the Global EM Et Sum for the entire Calorimeter - it is fed to 4 separate comparators, one for each L1 EM Et Ref Set. These Ref Sets are programmed by COOR, and also have a 0.25-Gev energy scale (and 8-bit resolution). The comparison results are fed into the 4 EM Et Counter Trees to produce the "number of Trigger Towers above Reference Et for EM Et Ref Set 3:0" counts used in L1 Triggering The output of the Had Et lookup memory by itself is also used in two ways: - it is driven to the Had Et Adder Tree to produce the Global Had Et Sum for the entire Calorimeter - it is fed to 4 separate Had Veto comparators, one for each L1 EM Et Ref Set. These Had Veto are programmed by COOR, and also have a 0.25-Gev energy scale (and 8-bit resolution). These comparators were not used in Run I. The output of the EM Et lookup memory is also ADDED to the output of the Had Et lookup memory to produce a 9-bit Total Et (again with 0.25 GeV energy scale) for each Trigger Tower. This 9-bit Total Et is used in two ways: - it is driven off the CTFE card as a 9-bit quantity and used to feed the Run I L1.5 Cal Trig and the Run II L2 Cal_pp's (note that we time-multiplexed TWO quantities -- EM Et and Tot Et -- on this single port. We accomplished this via "double cycling" the lookup memories to eliminate the Had contribution for a period of time. This "double cycling" must continue for Run II. See the next section for important details regarding these EM Et and Total Et signals. - the LSBit is DROPPED (giving 0.5 GeV energy scale "with truncation") and fed to 4, 8-bit comparators, one for each L1 Total Et Reference Set. The comparison results are fed into the 4 Total Et Counter Trees to produce the "number of Trigger Towers above Reference Et for Total Et Ref Set 3:0" counts used in L1 Triggering Separate from the addition described above, the EM and Had ADC Counts are also added together and fed to two additional lookup memories, called the Px and Py lookup memories. These memories do the following: - apply a symmetric low-energy cut (max of 0.5 GeV or 2.5 sigmas of electronics noise) - shift the fixed pedestal to a new value which is a function of eta (for purely technical reasons) - apply z-vertex correction (not used in Run I) - apply the cos(phi) or sin(phi) factor for Px and Py, respectively - they DO NOT modify the energy scale The output of these memories is used in two ways: - they are driven to the Px and Py Adder Trees to produce the Global Px and Py Sums for the entire Calorimeter (used to calculate Missing Pt) - using a "hook" built into the Px and Py Adder Trees, and again "double cycling" the Px and Py lookup memories, the Large Tile Et is calculated and compared to 8 Large Tile Reference Sets. These Reference Sets are programmed by COOR and have a 0.25 GeV energy scale). The comparison results are used to produce the "number of Large Tiles about Reference Et for Large Tile Ref Set 7:0" counts used in L1 Triggering It is important to note that the Large Tile Et is simply the sum of the EM and Had ADC Counts for 32 Trigger Towers (with no symmetric low-energy cut or vertex correction as applied to the Trigger Tower Total Et). For the Run I L1.5 Calorimeter Trigger -------------------------------------- See D0 Note 2117, pages 17-18. On the web at: http://www.pa.msu.edu:80/hep/d0/ftp/run1/l15/caltrig/d0_note_2117.txt Input data from the L1 Cal Trig was 9-bit Trigger Tower EM Et and Total Et as described above. Note that EM Et, although expressed as 9-bits here, is fundamentally only 8-bits. Only Total Et, the sum of 2 8-bit numbers, could actually have 9 meaningful bits. Between the L1 Cal Trig and the L1.5 Cal Trig, these 9-bit quantities were "squeezed" into 8-bits, in the following way: EM Et: Energy scale was 0.25 GeV per count (saturation around 60 GeV) Pedestal variable as function of eta, range of 16-22 counts NO symmetric low-energy cut i.e., except for the details of the pedestal, these are EXACTLY the Trigger Tower EM ADC Counts referred to in the above section ^^^^^^^^^^^^^ Tot Et: Energy scale was 0.25 GeV per count (saturation around 60 GeV) ^^^^^^^^ Pedestal variable as function of eta, range of 16-28 counts Symmetric low-energy cut, maximum of 0.5 GeV Et or 2.5 sigmas of electronics noise (applied separately to the EM Et and Had Et components) i.e., except for the energy scale, these are EXACTLY the Trigger Tower Total Et values described above. ^^^^^^^^ For the Run I L2 Trigger (and Offline and Verification and Simulator...) ------------------------------------------------------------------------ For Trigger Towers, the data provided by the L1 Cal Trig to L2 and beyond via the Data Block was the EM and Had ADC Counts referred to above. Any downstream system which wanted to reconstruct Trigger Tower EM Et and Total Et as seen in the L1 Calorimeter Trigger (and verify that the Reference Set comparisons were correctly done) could do the math and comparisons exactly as done in the L1 Cal Trig (i.e. they had to know the programming of the EM and Had lookup memories and know about the "truncation" at the input to the Total Et Ref Set comparators. For the Run II L1 Calorimeter Trigger ------------------------------------- The input data will remain the same, although there are questions regarding how well the signals from the BLS will actually characterize the single- crossing Trigger Tower energy at either the 396 or 132 ns crossing rate. The structure of the L1 Cal Trig will remain very similar to that used in Run I. We will have to modify the analog front end of each CTFE card (hopefully only the resistor modules), and re-tune the sampling time of each CTFE (which is set, unfortunately, via DIP switches rather than via a programmable register). We will also have to make new PROMs for each type of lookup memory on the CTFE cards. Before making these PROMs, we have many decisions to make: - will there be z-vertex correction? - if not, what will we use the multiple pages for? - or, should we already reserve some pages for other functions? - what are the details of the symmetric low energy cuts for Run II? - does 2nd lookup get these cuts? different cuts? - what is 2nd lookup used for besides Large Tiles (Px/Py) and feeding L2/L3 (EM/Had)? We will need to make the L1 Cal Trig actually run in pipelined mode. We also need to continue double-cycling the EM and Had lookups (and feed the time-multiplexed result to the ERPB cards) and the Px and Py lookups (note: the Px/Py double-cycling must continue through at least the tier 1 CAT2 cards). We need to provide "quadrant matching" outputs (actually, 4 "quadrant" outputs for each of Central Calorimeter, End Cap North, and End Cap South). There are multiple options for generating these signals. They should be based on the EM Et for either Trigger Tower or Large Tile sized areas being above a programmable Reference Et. There are requests for multiple "count thresholds" for the Reference Set, and possibly for multiple Reference Sets. The Cal Trig may internally have the possibility to "self-Veto" any of its And-Or Term outputs based on the state of (one? multiple, OR-ed together? how programmable?) other of its And-Or Term output(s). This would be used to counteract the effects of "sharing" energy between crossings due to the integration time of the BLS. Another technique to counteract this sharing (for the L2/L3 data only) might be to apply some digital signal processing to the Trigger Tower EM Et and Total Et on the ERPB. We would have to be very careful in deciding to do this, because we would not be providing an independent "uncooked" data set to L2/L3 (e.g. the Trigger Tower ADC Counts). For the Run II L2 Calorimeter Trigger (and Run II L3) ----------------------------------------------------- Again the input will be the 9-bit EM Et and Total Et described above. It will be again necessary to "squeeze" the 9-bits into 8-bits. For EM Et, there is no question but that it should remain at 0.25-GeV resolution. For Total Et, there are two obvious choices: - 0.25-GeV resolution (like the Run I L1.5 Cal Trig, and as described in my data format note of 24-Mar-1999) - 0.5-GeV resolution "with truncation" (like the Run I L1 Cal Trig, and as described in the TDR) Additionally, the question of symmetric low energy cuts must be addressed. How should they be handled for EM Et and Total Et data for the L2 Cal Trig (and L3)? We do not have infinite flexibility to select cuts used for L1 Triggering independently from cuts used for L2 and L3, but some freedom does exist. These choices should be driven by the physics. They need not be made immediately, but within say 6 months. Once made, they cannot change from run to run or week to week. In all cases, the EM Et and Total Et data will have uniform pedestals (no eta dependence as in the Run I L1.5 Cal Trig). Note that, for L3, the L1 Cal Trig will NO LONGER provide the EM and Had ADC Counts for each Trigger Tower. This is a fundamental change in the data being provided to L3. Finally, due to the potential for energy sharing between crossings mentioned above, L3 (and L2) may want Trigger Tower Et information for multiple crossings. The details of this have yet to be understood.