The first-level (LVL1) trigger works on a subset of information from the calorimeter and muon detectors. It requires about 2 micro-seconds to reach its decision, including the propagation delays on cables between the detector and the underground counting room where the trigger logic is housed. All of the information from the detector must be stored in pipeline memories until the LVL1 decision is available.
For events selected by the LVL1 trigger, the information from the detector must be retained for further analysis. The data for such events are transferred to readout buffers where they remain until the LVL2 decision is available. The data can be accessed selectively by the LVL2 trigger which uses regions of interest defined by the LVL1 trigger as indicated in the figure.
The LVL2 trigger refines the selection of candidate objects compared to LVL1, using full-granularity information from all detectors, including the inner tracker which is not used at LVL1. In this way, the rate can be reduced to ~1kHz. Many events are analysed concurrently by the LVL2 trigger system using processor farms, and an average latency of up to ~10ms is considered reasonable.