From stefano.piano@ts.infn.it Tue May 18 15:56:42 2004 Date: Wed, 05 May 2004 15:49:05 +0200 From: Stefano Piano To: finudagroup@lnf.infn.it Subject: Minutes of the FINUDA online meeting - Pavia, 29/04/2004 Minutes of the FINUDA online meeting Pavia, 29/04/2004 Present: Diego Faso faso@to.infn.it Valerio Filippini Valerio.Filippini@pv.infn.it Paola Gianotti paola.gianotti@lnf.infn.it Simonetta Marcello marcello@ph.unito.it Alberto Panzarasa alberto.panzarasa@pv.infn.it Stefano Piano stefano.piano@ts.infn.it Fulvio Pompili Fulvio.Pompili@lnf.infn.it Giuseppe Simonetti Giuseppe.Simonetti@ba.infn.it The following items were discussed: 1. DAQ stability: a. The new run control (FNDRUN) will be installed on the 10th of May 2004 (Gianotti, Piano) and the whole data acquisition will be tested in order to evaluate the stability of differential and optical PVIC system with the new run control. b. During the week from the 10th to the 14th of May 2004 the new 100 Mbit/s ethernet cards will be installed (Gianotti, Piano) on the CES diskless cpu's. Some tests will be performed to evaluate the possibility to change the cpu's to geb data transfer using Ethernet UDP packets instead of differential PVIC, which sometimes freezes run control. Valerio Filippini advised the online group about the OS latency time when UDP packets are received. 2. DAQ performance: a. The main sources of dead time in our acquisition system were analysed. They were the injection inhibit, the trips of the detectors (usually the drift chambers during injection) and the correspondent slow trip recover time, the run control freezing and the I/OSIM cram's to cpu's data transfer. i. To reduce the dead time it is crucial to improve the slow control interface: one click to recover the whole FINUDA is less time spending than to remember several directions. The mega console development is needed. Beside, it is crucial to send from mega console and slow controls to run control some information as the gas mixing and the detector current values in order to save this information into raw data. The collaboration will be advised about the doubt "who has to develop the slow controls". ii. Another crucial job is to improve the DAQ saturation rate due to I/OSIM cram's to cpu's data transfer. On the 11th of may 2004 the CAEN people will change the DMA chip, the memory and the clock of I/OSIM crams. The rate of DAQ saturation will be improved from 66 Hz to 120 Hz. In order to bring our saturation rate to 240 Hz we can split into 4 parts the I/OSIM VME crates (at present 2 parts). Before splitting VME crates an evaluation of I/OSIM DAQ program performance is needed (Filippini, Piano). Valerio Filippini told about the possibility to use assembler language to develop the CM noise calculation and to improve I/OSIM DAQ performance. 3. Packet size and raw data file size. The packet size during the last runs was about 33 KB/event that means that 20000 events = ~640 MB (~430 MB gunzipped). We cannot use larger file because the max file size with 32 bit pointer limit is 2 GB and it is very difficult to handle file with the size greater than 750 MB (one cd-rom). However, it is possible to reduce the packet size. In the raw data there are 8 spies x 8 chips x 3 hybrids x 18 modules = 3456 spies for each event coming from I/OSIM. They correspond to about 14 KB/event. There are two ways to follow: the former is to change the raw data format in order to reduce the size of spies, the latter is to reduce the number of the spy channels acquired (Filippini, Piano). 4. Acquisition of radio frequency signal (RF/4) as Common Stop signal to TDCs. First of all, it is not clear whether RF/4 signal is useful to have better TOF resolution for two reasons: 1) using RF/4 it is necessary in offline analysis to establish for each event the bunch crossing (among 4 bc) originating the event; 2) the final observables with physics meaning are the TOF differences between TOFINO and TOFONE (time of flight of particles), so even using a very sharp signal as RF, when calculating TOF differences what is relevant is the TOF resolution of TOFINO and TOFONE. That means the problems coming from a not sharp Common Stop signal as OR TOFINO is cancelled making TOF differences. In case the acquisition of RF is demonstrated to be needed, it is necessary to study an electronic module to acquire the RF/4 (or RF) signal (Pompili) and to send the RF/4 signal to TDC. 5. Further hardware upgrade: a. TOFINO cables (cross-talking trouble and mean-time) (Marcello) b. 2 TDC Caen to acquire high-threshold mean-time signals (Marcello, Gianotti) c. 24 channels NIM amplifier module for TOFINO signals (Pompili) d. Camac interface (at least one more) (Pompili) e. Prompt coincidence (Marcello) f. To be discussed after CRAMS upgrade: 2 diskless cpu's, 2 CORBO'S, 1 VME CRATE (the corbo modules are not produced anymore, maybe it is possible to look for them in closing experiments) g. Constant fraction and common front-end electronics: as already discussed in the last Collaboration Meeting at LNF, it is not possible to change the CFD modules in a short time as 1 year, because it is necessary to design and realise a new module (Pompili). At the moment such a design phase cannot start since it is not sure which TOFINO will be installed (because old TOFINO with HPD gives slow signals and new TOFINO with PM gives fast signals). 6. Monitor and pre-analysis software: a. some improvements are needed to better understanding the detector behaviour during the data taking (Cerello): i. one display with the summary status for each detector ii. a print-out at end of the run iii. bell alarm if one or more histograms are not filled (check the histogram integral) iv. one refresh button to refresh all histograms v. re-binning histograms without recompiling froot vi. more stability during the data taking: aa. avoid as more as possible geb2hdt crash bb. capable of surviving at the end and the restart of the run b. FINUDA needs to publish some pre-analysis results on the web pages (Simonetti), the online web page will be moved from monitortof to fnddisk01 to display: i. luminosity (bhabha, kaons and counter methods) ii. beams collision positions iii. boost calculations 7. Data: a. electronic logbook (Simonetti) b. automatic backup procedure (Piano) c. next storage system for the Finuda experiment: i. before the next data taking: the data will be stored at LNF data storage system ii. before the next data taking we must delete the data of fnddisk01 iii. next year (February) it will be possible to estimate the data occupancy and to buy EIDE disks to improve fnddisk01 capacity d. Elonex asks for substitution of the fnddisk01 disks due to timeout problem (Piano) 8. Trigger System: a. Next run timing gate to TOFONE from 20 ns to 10 ns (Marcello) b. Prompt coincidence substitution (Marcello) c. Trigger timing: for the next data taking it is needed an unambiguous timing made by laser. During the last data taking the final timing between TOFINO and TOFONE was done with the beam collisions because the optical fibres were not "safe calibrated". It seems possible (Marcello) to measure optical fibre length, or better the length difference between TOFINO and TOFONE, using cosmic ray data. The TDC-data coming from TOFINO are affected by the CFD faulty operation (Faso) and maybe also using cosmic rays it will be difficult to find the right time alignment. In any case the events clearly classified as faulty can be cut instead to be recovered (Marcello-Gianotti).It is clear that FINUDA needs a time alignment task-force. d. A second level trigger in FINUDA can be done only using signals from TOF, because the use of I/OSIM signals (which are the most interesting to perform better selection of events) is not possible at second level (Marcello). A second level trigger using signals from different detectors need a new FEE. Filippini/Panzarasa have studied the possibility to implement a second level trigger using TOF, which performs a better selection of events based on Cluster/Topology. Such a trigger has the advantage to be realised fully by software, computing the TOFONE clusters after TOF event readout with an expected latency of about 1 ms. It is possible to improve our acquisition rate stopping the I/OSIM acquisition if the events are not good. This proposal will be shown to the collaboration during the next collaboration meeting (Filippini). 9. To buy: a. TOFINO cables (cross-talking trouble and mean-time) (Marcello) b. 2 TDC CAEN to acquire high-threshold mean-time signals (Marcello, Gianotti) c. 24 NIM amplifier (Pompili) d. CAMAC interface (at least one more)(Pompili) e. Prompt coincidence (Marcello) f. To be discussed after CRAMS upgrade: i. 2 diskless cpu's ii. 2 CORBO'S (the corbo modules are not produced more, maybe it is possible to find them in some laboratories or old experiments...) iii. 1 VME CRATE g. Next Year: i. Disks (LNF afs ~1.5 TB) ii. Fnddisk01 EIDE disks (to be evaluated)