This upgrade approach is based on the Pulsar board, a general purpose VME board developed at CDF and used for upgrading both the L2 global decision crate and the L2 silicon vertex tracking (SVT) subsystem. In this approach, the full resolution calorimeter trigger tower data would be received, preprocessed and merged by a set of Pulsar boards before being sent to the Level-2 decision CPU running a cluster-finding algorithm.

The challenges of this upgrade are to keep the clustering algorithm processing latency within ~20 microseconds, and to have minimal impact of the running experiment during commissioning. Since the actual cluster-finding is done inside the CPU, it is much more flexible. The clustering algorithm is more robust against increasing luminosity (higher occupancy in the calorimeter system). For example, the large fake clusters with large fake ET will be eliminated since the CPU algorithm will not blindly combine contiguous regions of trigger towers. In addition, the event SumET and MET can be re-calculated using the full resolution 10-bit trigger tower energy information available to L2. The proposed L2CAL upgrade allows more sophisticated algorithms to be implemented.

Jet reconstruction using a cone algorithm which is currently being done at Level 3 can be moved to Level 2, albeit clustering trigger towers (instead of physical towers) and using only a single iteration in order to save processing time. We will show that L2 jets found using the proposed algorithm are nearly equivalent to offline jets in terms of ET , centroid, and efficiency, a vast improvement over the current situation. The calculation of MET at Level 2 with resolution closer to that at Level 3 will reduce L2 rates for Higgs and SUSY triggers that require ET , which will be vital for preserving these triggers at high instantaneous luminosity.

Although the main goal of the upgrade is to significantly reduce the growth terms of the existing jet and MET related triggers, the proposed L2CAL upgrade for CDF will make the full-resolution trigger tower information directly available to the L2 decision CPU. This is a big step forward in improving the CDF triggering capability at Level 2. There are two aspects: to have enough flexibility to deal with potential new challenges at the highest luminosities, and to improve CDF new physics search sensitivities beyond the baseline. For example, the new system would allow the possibility to trigger on dijet mass, Δφ between dijets or between a jet and ET , sum ET of the clusters (HT), and better jet-SVT matching for b-jets. In addition, together with the XFT upgrade which makes ~3D tracking information available at Level 2, this might provide a means to improve important tau triggers at Level 2.