Trigger

A level 0 trigger operates synchronously with the storage ring, the decision being made in the 360 ns between crossings of the seven beam bunches. In this way, failed level 0 triggers incur no dead-time. At 5.28 GeV and , the level 0 trigger reduces the 2.8 MHz beam crossing frequency by a factor of 300 to 10 kHz. Examples of existing level 0 triggers are:

• Any two separated barrel time of flight counters
• Opposite endcap time of flight counters
• One time of flight counter and a track segment in the vertex detector
• One high threshold barrel crystal region
• Opposite high threshold endcap crystal regions

A limitation of the present system is that mixed tracking and calorimeter triggers can not occur at this level because of their disparate signal development times. CsI energy trigger information is available after 2 s. The self-resetting feature of the drift chamber electronics causes the wire hit information to be reset 800 ns after the discriminator fires if no trigger inhibits the reset.

Successful level 0 triggers incur experimental dead-time to freeze the contents of the front-end electronics and initiate the level 1 trigger hardware which has a decision time of 1.2 s. Failed level 1 triggers entail another 1 s dead-time to reset the front-end electronics. At 5.28 GeV and , the level 1 trigger reduces the 10 kHz level 0 trigger output by a factor of 300 to 30 Hz. This includes pre-scaling of QED processes in the endcap regions. Some examples of level 1 triggers in use today are:

• Back-to-back high threshold crystal regions in the barrel or in the endcap
• One time of flight counter, two track segments in the drift chamber, and one minimum ionizing crystal region
• One each high threshold barrel and endcap crystal regions

Successful level 1 triggers initiate, if appropriate, the level 2 trigger which consists of two hardware tracking processors. One uses the inner 10 layer drift chamber (VD) and the other uses the axial layers of the main drift chamber. One works with digital hit information formed from the OR of 4 drift cells. The other uses wire-level hit information to find tracks in predefined road-widths. The total processing time is 40 s. Under 1994 running conditions, this trigger rejected 50%of the level 1 triggers.

Accepted level 2 triggers initiate front-end readout by the data acquisition system where the data is buffered at the input to the VME-based event builder system. The trigger is then re-enabled and the data sparsification and event building proceed asynchronously. After event building, data are transferred to an on-line workstation where they are analyzed by the level 3 software trigger. Access to the full data allows this code to calculate such quantities as neutral energy balance and to use the inner tracking detector hits to eliminate beam-wall events. Level 3 reduces the trigger rate by about a factor of two.

The overall performance of the trigger chain can be simply determined by examining the fraction of events that are qualified as ``good'' by the off-line reconstruction code. Fifty percent of the events processed by the level 3 trigger are rejected. They are generally categorized as not containing charged tracks or not being interesting pure neutral events. The efficiency of the trigger system has been determined to be %for events and nearly 100%for hadronic events.