December 13, 2006 online meeting minutes
presented: Sergey Boyarinov, David Doughty, Mark Ito, Elliott Wolin, Vardan Gyurjyan, Sergey Pozdnyakov, Nerses Gevorgyan, Ed Jastrzembski
We had very useful discussion on CLAS12 trigger design. The project
was discussed already on a meeting few days before involving Volker
and Latifa so we had already general directions and Sergey prepared
preliminary document (see attachment). Today we are concentrated on
several issues which were not completely understood. One of them was
how we can search for clusters in calorimeter(s) using JLAB Flash
ADC SUMs as trigger system inputs. It seems possible to identify
cluster position but unclear how to obtain cluster energy. We decided
to put coordinate-based cluster finding into design along with energy
sum for predefined parts of calorimeter(s), and keep cluster energy
finding as an option. Several other aspects were discussed, such as
the possibility to use coincidence-based logic instead of
latching-based, usage of SYNC and ASYNC inputs as it is done in
current trigger system, missing trigger bits in TS latch register for
multilevel triggers etc. More meetings will be held to clarify
remaining details.
G13 run was shortly discussed. Our time profiling system identified
TAGE2 ROC as the current bottleneck. New v1190/v1290 CAEN TDC
firmware conveniently arrived and was loaded into TAGE2 ROC TDCs, and
readout library was optimized as well. As result we were able to
increase event rate from 8kHz to more then 9kHz at dead time around
15% (can be viewed using online timeline plots).
Elliott fixed logbook acknowledgments and run_summary script.
We also talked about SY1527 incorporation into EPICS and about
counting room test setup for old Lecroy HV mainframes. Both projects
goes well: we have medm GUIs working, new classc0 IOC was installed
in counting house along with one Lecroy mainframe, and Delisa is
working to fix 4 more LECROY mainframes. classc0 will be used to test
Lecroy mainframes and HV boards before installing them to the Hall.
Nerses is working to produce documentation for SY1527
incorporation into new EPICS and on Linux IOC in general.
Classc0 is running old EPICS. It maybe useful to try to convert
classc0 to new EPICS to identify corresponding problems.
Attachment: CLAS12 trigger project: short-term plan (Dec 2006-May 2007)
Overview
The goal for 6 following month is to develop principal scheme of trigger system for June 2007 CD2 review. Plan includes 3 steps: preliminary design (Dec 2006), CLAS data- and simulation-based efficiency studies (Mar 2007), electronics design (May 2007).
Preliminary design
CLAS12 trigger system consists 3 levels discussed below: Level 1 is based on information from 6 detectors: forward and preshower calorimeters, forward and central time-of-flight counters, high- and low-threshold cherenkov counters. Level 1 must have a possibility to include additional detectors during CLAS12 operation if needed. Three kind of information will be used in Level 1 trigger: time, position and energy. Time information will be enforced by the nature of trigger logic which is based on coincidence strobe with 100+ MHz frequency. Position and energy can be obtained from pretrigger discriminators as it was done in CLAS, or/and from Flash ADC boards.
Level 1 decision is based on energy cuts for pre-defined objects, geometry cuts for position-based reconstructed clusters and (optionally) energy cuts for reconstructed clusters. Matching of different components will be used as well to set additional cuts and provide extra information for matching with Level 2 trigger. Decision time - tens of nanoseconds.
Level 2 is based on drift chamber hit information. It will perform superlayer- or region-based segment finding and sector-based road finding. Since existing ADB crates will be reused they will provide an information for the replacement of old segment collectors and following logic. Level 2 decision is based on road multiplicity cuts and matching with Level 1. Decision time - few microseconds.
Level 3 is based on information from all CLAS12 components obtained from DAQ and slow control systems. It will perform event rate and event size reduction. We assume that full event reconstruction using default geometry and calibration will be incorporated. Preliminary design must be ready by the end of 2006. All three level details will be finalized as result of CLAS data analysis to be performed in three following month.
CLAS data- and simulation-based efficiency studies
It is very important to estimate an efficiency of every proposed trigger component. Following components must be studied: - electron identification efficiency using cluster finding in both calorimeters (coordinate- and energy-based) in conjunction with hits in time-of-flight and cherenkov counters; - electron identification with and without proposed Level 2 information; - possible trigger efficiency improvement for experiments without electron identification and for experiments with multi-particle requirements. All efficiency studies must be completed by the end of March 2007.
Electronics design
Electronics design discussion was started and will continue until March 2007, then actual principal scheme development will start. It must be ready by the end of May 2007.