The bunch crossing (BC) frequency at HERA accelerator is approximately 10 MHz (96 ns — bunch crossing time), and each bunch collision produces some signal in the detector with synchrotron radiation and photoproduction, apart from the signals which come from beam halo, cosmic muons and electronics noise. The rate of theep scattering at highQ2 is about 104 times smaller. At the same time the manageable rate for the H1 data acquisition system is about 40 Hz.
Therefore it is essential to trigger on the collisions of interest. For this purpose, the H1 detector uses a pipelined multi-layered trigger system schematically shown in Fig. 4.5. The filtered data from one level are sent to the next level until manageable rate is reached. Levels L1, L2, L3 are hardware triggers, level L4/L5 are based on the online reconstruction of events.
4.3.1
The First Trigger Level (L1)
The first trigger level (L1) makes a decision within 2.3 µs on whether to accept or reject an event using the information provided by the different L1 trigger el- ements (TE) — signals sent by various subdetectors of the trigger system. The central trigger logic (CTL) combines these TE into 128 subtriggers. The response time (needed for charge collection, transportation of the signal from subdetector to CTL, etc.) for some subdetectors is relatively large, therefore the information fist is sent into pipeline where it is kept for 24 BC (about 2 µs). Then the TE are linked logically and the L1 decision is made. If any of the subtrigger conditions is fulfilled, the signal from the event is passed to the next trigger level. At the same time the pipeline is stopped and the dead-time of the detector starts.
The subtrigger could be prescaled. If the prescaling of the subtrigger is p, than only each p-th positive decision for this subtrigger can trigger the positive decision for the event. The prescaling is used for the subtriggers triggering relatively common
4.3. TRIGGER SYSTEM 45
physics or used for performance of the detector to balance rate of such events with rare physics events. The L1 decision for the subtrigger X before the prescaling condition is applied is called raw and actual otherwise.
The L1 trigger elements essential for the analysis are the following
The LAr Calorimeter Trigger Elements
The LAr calorimeter provides signals to the CTL [53]. The analog signals provided from 45000 LAr cells added to “trigger cells” (TC) and further to “trigger tow- ers” (TT) digitized using FADCs (flash analog to digital converter) and summed into 256 “big towers” (BT), having a projective geometry relatively to the nominal interaction point, see Fig. 4.6. Energy deposited in hadronic and electromagnetic parts of the calorimeter are counted separately. Several thresholds are introduced to suppress noise and backgrounds, there are so-called AGM-threshold for the sum of the analog signal, and the BT-threshold for the digital signals from BTs.
Figure 4.6: The projective Big Towers in LAr calorimeter. Each tower points toward the nominal interaction point.
The LAr TE used in the analysis are:
• LAr electron 1 and LAr electron 2. The TEs are set if the electromag-
netic energy deposited in the BTs exceeds a specific for this TE threshold (see Fig. 4.7) and the energy in the associated hadronic big tower is lower than a certain threshold value.
• LAr T0(or “event T0”). The TE is set when the number of the BTs giving a
T0 signal exceeds a certain value. For the actual setup one BT T0 is sufficient.
The Track Trigger Elements
The tracking chambers CJC, CIP, COP, COZ all provide the information for the CTL decision, details can be found in [50]. The trigger elements used in the analysis are
• CIP T0. The TE set if there is at least one central track in coincidence with
the interaction time.
• CIP mul and CIP sig. See below.
4.3. TRIGGER SYSTEM 47
Figure 4.7: The thresholds for the LAr electron 1 and LAr electron 2 trigger elements as a function of Θ.
The Veto Conditions
Veto conditions are applied to most subtriggers in order to reject the backgrounds from beam halo and beam-gas interactions.
• ToF Veto. The condition is used to reject the background events on the basis
of the Time-of-Flight information. These trigger elements are based on the VETO BG, BToF BG, SToF BG, and FIT trigger elements of the ToF system described in Section 4.4.
• CIP Veto. The CIP information is used to reject the background from the
collimators close to the H1 interaction region. The requirement on the CIP: (CIP mul > 7) && (CIP sig == 0) efficiently rejects the background events with a high track multiplicity.
The SpaCal Calorimeter Subtriggers
The signal from the localized energy deposition in the electromagnetic section of the SpaCal calorimeter is also used as a trigger; in case of NC at high Q2 it reacts on the hadrons which go to the backward direction. The SpaCal subtriggers involved are s0 (threshold energy is Ethr = 6 GeV, minimal cluster radius is rsp = 12 cm) and s9 (Ethr = 2.5 GeV, r
sp = 15 cm). To keep a reasonable rate at low Ethr, s9 contains an extra track condition.
4.3.2
The Second/Third Trigger Level (L2/L3)
The second level trigger exploits two trigger systems — the Neural Network Trigger (L2NN) [54] and the Topological Trigger (L2TT) [55] — and makes a decision within 20 µs. At this level the full information from all subdetectors is available, and there is enough time to inspect the correlations between them.
The results of the L2 trigger are given to the central trigger L2 decision logic (CTL2).
For the analysis theL2TT LAr electronTE is used. The condition is based on the approach in which the detector is represented with a matrix in Θ−φcoordinates. This matrix is used to find a “distance to the background” and make a trigger decision. The detailed description of the level 2 trigger can be found at [56].
The third level trigger was not active until the end of 2005 year and was not used for this analysis.
4.3.3
The Fourth/Fifth Trigger Level (L4/L5)
The trigger levels L4/L5 performs the event full reconstruction from the complete readout of the event information [57]. It is the software asynchronous trigger. Event parameters (tracks, clusters, vertex) are reconstructed by the H1 software algorithms (H1REC, see below) on the dedicated PC farm.
The events are classified then with the so-called L4 filters. For the classes where statistics is enough only part of the data is stored and the rest gets “L4 trigger weights” applied to the data events. Events do not fit any interesting class are rejected.
The raw date for the events surviving all the trigger requirements are written to the production output tapes (POT, about 100 kb per event) and the reconstruction information is stored as the data summary tapes (DST, about 10 kb per event) at a rate of about 10 events per second.