CMS Bulletin

TRIGGER

Level-1 Trigger Hardware and Software

The overall status of the L1 trigger has been excellent and the running efficiency has been high during physics fills. The timing is good to about 1%. The fine-tuning of the time synchronization of muon triggers is ongoing and will be completed after more than 10 nb-1 of data have been recorded. The CSC trigger primitive and RPC trigger timing have been refined. A new configuration for the CSC Track Finder featured modified beam halo cuts and improved ghost cancellation logic. More direct control was provided for the DT opto-receivers. New RPC Cosmic Trigger (RBC/TTU) trigger algorithms were enabled for collision runs. There is further work planned during the next technical stop to investigate a few of the links from the ECAL to the Regional Calorimeter Trigger (RCT). New firmware and a new configuration to handle trigger rate spikes in the ECAL barrel are also being tested. A board newly developed by the tracker group (ReTRI) has been installed and activated to block resonant trigger frequencies that could possibly damage wire-bonds on CMS detector ICs. The algorithm used was developed by CDF. A batch of new central Trigger Control System modules has been produced whose firmware can be programmed through VME. The parameter settings of the beam position monitoring (BPTX) trigger electronics have been optimized.

Before the start of LHC run in 2010 various other improvement to the online software and firmware were made, in particular: the luminosity section was shortened to 1018 orbits (~23 sec); a mechanism to automatically mask subsystems from L1 configuration using FED vector was put in place; automated resynchronizations in response to out-of-sync signals from detectors were enabled; new GCT firmware implemented an improved tau algorithm; the configuration of LHC clock changes and ramping was automated; the monitoring of LHC clock and orbit was improved; and the trigger shifter environment was enhanced.

Level-1 Trigger Commissioning and Operations

At the re-start of LHC operations the Beam Pickup and Beam Scintillator triggers were re-synchronized. This allowed successful delivery of first 7 TeV collisions with precise trigger timing at the Media Event.  The next LHC fills were used to perform clock phase scans for the HF, ECAL and CSC systems while relying on Beam Scintillation Counter as the main minimum bias trigger. Results from these scans contributed to improvements in the synchronization of Trigger Primitives from these subdetectors. Subsequently, all Level-1 object triggers have reached the condition of less then 1% of out-of-time triggers and could be switched from monitoring mode to actively trigger the CMS data acquisition. The final component to complete the fine timing process is the barrel muon trigger, which awaits increased luminosity. Needless to say, more statistics will also allow setting the timing of all individual trigger channels to maximum precision.

An important part of Level-1 Trigger commissioning with collisions is the study of efficiencies. Despite of the lack of a proper signal, especially at higher pT or ET, reconstructed signal-like objects triggered by minimum bias triggers are used to identify various sources of inefficiency.  This helps to tune various hardware parameters, e.g. Trigger Primitive thresholds or timing of inter-channel communication, and to apply and test the first corrections on Level-1 threshold quantities.

Despite the rapidly evolving LHC conditions, the smoothness of the operation of the Trigger has improved over time. The scope of the Trigger Shifter activity moved from taking care of configuration to monitoring the Trigger performance. The automation of trigger configuration tasks, one of the most recent being the automatic clock source selection, brought a reduction of configuration errors and contributed to an overall high level of data taking efficiency. The main responsibility to maintain correct configurations and keep continuity resides in hands of experienced Trigger Field Managers who rotate on a weekly basis. The newly introduced Offline Trigger Shifts ensure proper monitoring based on Offline DQM. The process of data certification is being improved with the help of extensive use of the Run Registry utility by all Trigger shifters.

Trigger Studies Group

It has been a very busy and productive period for the Trigger Studies Group (TSG). Since the end of March the instantaneous luminosity delivered by the LHC has increased by two orders of magnitude, up to about 21029. Several trigger menus were prepared, validated and deployed in the Filter Farm. The trigger performance has been monitored constantly and CMS has taken data with good quality triggers.

Trigger menus have been developed for every major luminosity milestone: 11027, 11028, 11029. The CMS trigger has gradually moved from the startup period with several commissioning triggers and a very open physics trigger, to a more constrained online selection with simplified minimum bias triggers and all physics HLT paths still running unprescaled. The triggered events are grouped into eight primary datasets to facilitate their timely reconstruction, with the addition of a dedicated Event Display stream for monitoring.

The trigger performance has been monitored at several levels:

• On the DQM side, with trigger object-based selection provided by the Physics Groups
• During data certification, to determine the quality of triggers in the data taken
• With collision data skims that are used extensively by the trigger experts to study the L1 and HLT behavior
• At the weekly Monday Trigger Performance meetings with regular reports from Level-1 and Physics Object experts
• By studying the CPU-performance of the HLT paths and providing skims of "slow events" to HLT developers.

The TSG has also been providing support for three software releases in parallel with daily work on:

• maintenance of multiple trigger menus
• release validation to identify areas where improvements can be made and determine which software release should be used for data taking
• MC production, which in the recent months has been using one of the latest online trigger menus

Operationally, a small dedicated team of HLT on-call experts has been ensuring the smooth and stable running of the CMS trigger. They not only drive the online deployment of the trigger menus, but also help collect and analyze error events and core files, producing more robust HLT software. This is an area where additional manpower is very welcome and can have a big impact on the quality of data collected by CMS.

Preparing for the next months, the TSG is developing new menus for every factor of two increase in luminosity. Rate predictions are extrapolated from recent runs with careful examination of contributions from collisions, noise and cosmics (see Figure 1). The effect of pileup on the CPU performance and efficiency of the HLT is being actively studied. Finally, work is being invested in a software tool that predicts individual trigger rates to guide the online shifter in detecting anomalies in real time.




by by Wesley Smith, from contributions of C.Leonidopoulos, I.Mikulec, E.Perez, J.Varela and C.Wulz