Minutes of the CALCOM meeting, January 30, 1998. ================================================ Agenda: D. Cords - On-line/DAQ progress L.C. Smith - EC energy calibration status J. Mueller - Tracking efficiency studies B. Niczyporuk - DC alignment and other issues ===================================================================== !!!! NOTE: There will be no CALCOM meeting on Friday February 6. !!! ===================================================================== Summary: ------- Several issues of relevance for the upcoming e1 and g1 runs were discussed. (1)The DAQ speed and event rate are, of course, of crucial importance for all data taking runs with CLAS. Dieter Cords presented, as a results of Steve Barrow's study, what factors are important for the current limitation in data rate and event rate. We may expect significant improvements if the number of read out ADC/TDC channels in the forward carriage can be reduced. This may be possible if the TDC discriminator and ADC sparcification threshold can be raised (which are currently set very low in the case of the TOF counters). (2) Drift chamnber calibration issues are now cleary coming to the forefront. Inclusion of the survey data improved the previously observed non-uniformity in the elastic ep peak versus azimuthal angle. However the elastic paek is still shifted by about 25-30 MeV in the RECSIS analysis. The position resolution also needs improvement. After the meeting a small working group was formed to address DC calibration issues. Bogdan presented his analysis using the ideal DC geometry and six sector symmetry. He finds variations from sector to sector for the elastic peak of up to 25 MeV, while the average is closer to the nominal value. He also sees sector=-to-sector variations in the reconstructed target position, as well as differences for inbending and outbending tracks. Jim Mueller presented results of a study to understand the tracking efficiency as a function of the luminosity. This will be crucial for defining the luminosity used in the upcoming 4 GeV running period. The (hit-based) tracking efficiency at a luminosity of 5.5x10^34/cm2/sec appears to be 97%, about one percent lower than at very low luminosity. It should be said that the exact origin of this improvement compared to older studies is not exactly clear (at least to me). Cole Smith reported on progress in the EC energy calibration. He includes all materials in front of the calorimeter in GSIM (exception CC) and gets a better (so not perfect) match with the reconstructed track momentum at lower particle momenta. There is still a difference in the absolute energy scale. However, it is not clear whether the cosmic ray calibration should actually give the correct absolute energy scale for electromagnetic showering particles. Volker Burkert ------------------------------------------------------------------- Individual reports: ================== D. Cords: On-line/DAQ Progress -------- The event rate of the DAQ System was tested with a 10 KHz pulser. Steve Barrow summarized his findings in 4 logbook entries under the system heading 'coda' (1/25 & 1/28). With an empty detector (HV off) and 40 ADCs always 'on' in the SC crate he achieved an event rate of 1.3 KHz. Vladimir Serov looked at data with raw banks from last December. He played the raw data back into the readout controllers and timed the secondary readout list. He finds .15-.2 ms/evt for DC2 and .2-.3 ms/evt for SC data. There is a problem with spurious signals in the data which slow down the 2nd ROL. The ADC pedestal offset should be increased by an amount to be discussed. A procedure was developed by Vardan Gyurjyan to monitor pedestals and to load sparsification thresholds. One can compare new and old pedestals (loaded thresholds) and look at fluctuations (standard deviations) and distributions. One can also turn off hot channels. This should give us a better handle on spurious ADC signals. L.C. Smith: EC energy calibration status ---------- Calibration of the forward calorimeter ADCs is based on analysis of MIP cosmic-ray tracks. Analysis of electron events of known momentum or pairs of photons from pi-zero decay can provide an absolute check on this calibration. Also the measured sampling fraction can be checked against monte carlo predictions. Results so far using pi-zeros show invariant mass peaks are within a few percent of the expected value for all sectors, using sampling fractions consistent with GSIM predictions. The average photon energy is 0.5 GeV in this comparison. For electrons, a systematic energy dependence is observed for the ratio of reconstructed EC energy to reconstructed DC momentum in all sectors. The quantity E/p-1 ranges from -10% at 0.5 GeV to +1-3% around 2 GeV. Since electrons are subject to ionization losses, an estimate was made of the energy lost by electrons prior to entering the calorimeter. The result was that 21-22 MeV is lost in the TOF bars, TOF backing material and EC cover plate alone. GSIM runs were performed to check this, and analyzed with RECSIS with energy cuts consistent with the ADC sparcification thresholds. The simulated energy response shows a decrease in E/p consistent with ionization losses by the electron in the above materials above 1 GeV. Below 1 GeV GSIM/RECSIS predicts a larger decrease in E/p than expected from ionization losses alone. However, the GSIM result is consistent with the decrease observed in the data. Currently studies are underway to study the effect of RECSIS/EC strip energy cuts on the low-energy response of EC. Details at: http://apollo.phys.virginia.edu/clas/ecsim.html J. Mueller: Hit Based Tracking Efficiencies --------- I gave more details on the HBT efficiency analysis presented at the Colaboration meeting. The idea is to find events consistent with an elastic collision, without requiring an electron to be found. Then one looks to see if the electron was found by tracking. The results are then compared for runs with different beam currents to see if there is any large luminosity dependence of the tracking efficiencies in the forward region. The analysis proceeds as follows: 1) Select positive track and assign them the proton mass. Calculate the missing energy and momentum recoiling against the putative proton. 2) Require that the sector opposite the "proton" includes a) a CC bank and b) the largest energy cluster in the ECHB bank have an energy (uncorrected for sampling fraction) > 20% of the missing momentum. 3) Require that the missing mass squared recoiling against the proton be within 0.1 GeV^2 of zero Latifa suggested at the Collaboration meeting that there might be some pion contamination in the "elastic protons" which would lead to a false inefficiency. A hand scan of the failures did show some inelastic events left in the sample. Perhaps this could be rejected by using the dE/dX in the SC. I implemented this, but although there are a lot of pions after step 1 above, it is reduced greatly by step 2 and effectively eliminated by step 3. However, I left in the cut, with the value set below the bottom of the observed proton band. 4) Require that the energy reported in the SCR bank for the TOF paddle matched to the proton be greater than 10 MeV. 5) The remaining events are binned as a function of the predicted polar angle of the "missing electron" 6) Require that a negative track be found within 3 degrees of the predicted polar angle. Make plot 5) again. Compare the plots to calculate efficiency as a function of polar angle. Although Latifa's suggestion turned out to be not required, it did cause me to re-run one of the jobs. The EC reconstruction seems to have improved in the last month, and the event sample seems cleaner after step 2 above. This caused all calculated efficiencies to improve. Below I present new numbers for the runs presented at the collaboration meeting. (only run 8055 was actually shown on Friday. The rest of the number are as of Saturday. The hit multiplicity numbers are from Steve Dytman.) Multiplicity SECTOR 1 SECTOR 1 RUN E_beam I_beam Region 1 HBT Eff.(OLD) HBT Eff.(NEW) 7730 1.6 1 nA 6.6 96.47 +- 0.39 97.84 +- 0.24 7731 1.6 2 nA 96.19 +- 0.39 97.40 +- 0.26 7744 1.6 3 nA 16.6 95.51 +- 0.59 96.47 +- 0.42 7745 1.6 4 nA 94.86 +- 0.44 96.56 +- 0.28 8055 2.4 5.5 nA 93.94 +- 0.87 97.03 +- 0.60 A hand scan of the raw data for the 24 failures in sector 1 run 8055, gives the following breakdown. no track by eye in sector 2 Spurious failure obvious inelastic event 1 Spurious failure extra segments in R1 or R3: track missed 5 track mis-reconstructed 11 Need to check TBT: maybe fixed Track found, looks good by eye, good chi2, but not within 3 deg of pred. 5 Spurious failure ? BOTTOM LINE: 1) It appears that even for the highest luminosity run I have looked at, the efficiency for FINDING an electron is >99%. 2) Mis-reconstruction affects 1-2 % of the found electrons at hit-based level. This is probably luminosity dependent, but I haven't quantified it yet. 3) I can easily think of two changes to the code which should improve the problem events. The time scale is probably on the order of a week to make, and test those changes to make sure I don't screw up something else. 4) I need to extend this study to measure resolution (mis-reconstruction) as a function of luminosity for time-based tracking. The current level of our tracking calibrations may limit our ability to see an effect at this time. Before the recent geometry changes, this was definitely the case. SHORT STATUS of EVENT RECONSTRUCTION and SIMULATION with SDA ============================================================ (Bogdan Niczyporuk, Jan.30,1998) SOME RESULTS (TBT) ================== Speed: about 100 trig/second on HP (jlabh3) DST contains: - about (12 - 35)% of original triggers (at least electron) - event size about 10% of the original event size with the following BANKS: EVTB(new) and original {CC,SC,EC,CALL} banks Calibrations ------------ The SDA use only the pedestals from "MAP manager" (SC,EC). Other calibration constants are derived from the data runs. In major, the success of time based tracking (TBT) depends on the knowledge of delays for DC (each signal wire) and for SC (each phototube). Those delays can be derived from data runs taken with "e1_EC_123456" trigger. At present, the current (preliminary) resolutions for the DC delays are about 10 ns and for the SC delays is about 2-3 ns. I have shown that the resolutions for SC delays can be easy reduced to the 1 ns level (work in progress). The resolutions of individual SC phototubes (Panel 1) are about (0.20 - 0.25) ns without time walk corrections. That is sufficient to see the RF beam structure. The SC delays for the Panel 2 and 3 were derived from the exclusive elastic and inelastic reactions. The DC delays derived from "pulser" runs are not complete and in the big portion are not applicable for the December runs (many adjustments has been made). Even using the pulser delays for the September runs produces worse DC resolutions. The 1 ns resolution for SC phototube delays seems to be sufficient for the successful completion of the time based tracking (to derive the particle velocity). DC Alignments ------------- At present the SDA uses the nominal geometry (design values). Deviations from the design values has been clearly observed. Analysis of elastic scattering (phi acollinearity of opposite sectors) suggests that the three pairs of sectors are misaligned by about 1 degree (see Table 1), but sectors 1 and 4 are acollinear by about 2.4 degree. From analysis of electron tracks with empty target (Run 7889, trigger e1_CCEC_1456) one can clearly see the walls of the target seperated by 4 cm. Following we show the z-positions of the downstream target wall and the distances to the z-axis (Yt) for electrons reconstructed at the different sectors: Sector#1 Zwall = 1.10 +/- 0.016 cm Yt = -2.10 +/- 0.036 cm Sector#4 Zwall = 0.63 +/- 0.018 cm Yt = 0.28 +/- 0.031 cm Sector#5 Zwall = 0.46 +/- 0.025 cm Yt = -0.74 +/- 0.051 cm Sector#6 Zwall = 0.43 +/- 0.023 cm Yt = -1.57 +/- 0.038 cm Again, the vertex positions data for sectors 1 and 4 seems to be correlated with the observed acollinearity. Track Reconstruction Efficiency (TBT) ------------------------------------- To obtain the hit based tracking efficiency is a simple task; one needs only the list of "dead" wires. The time based tracking efficiency is much more complex procedure since that evolves not only DC but also other subsystems (geometry misalignments, resolutions, time delays, drift cell physics, etc). In order to get a feeling how efficient is the TBT I have analyzed a few December runs taken with the "e1_EC_123456" trigger. As a denominator I have used the identified electrons with the DC (HBT), SC (slab matching) and EC (U,V,W matching and ratio Ecluster/p). A number of survived electrons (DC, SC timing and global fiting) gives the estimate of efficiency. The efficiency, averaged over 6 sectors, is about 96% which is only about 2% lower what the MC simulation predicts. Overall Resolution ------------------ The overall detector resolution can be judged by examining the width of elastic peak (ep --> e + X) and the misssing mass width of neutron (ep --> e + pi + X). The resolution (sigma) of elastic peak is about 15 MeV and the position is is about 945 MeV. The SDA simulation (by definition: perfect alignments and delays are ideal) and reconstruction (the same code) gives the sigma of 12 MeV and perfect position. The analysis of the two prong events (ep --> e + pi + X) gives the neutron mass about 945 MeV and sigma about 15 MeV. The above quoted values are for the December runs (Ebeam = 1.645 GeV, Itorus = 2250A) and were averaged over the all sectors. Table 1: December Run#7841 with trigger "e1_EC_123456" and Ebeam = 1.6 GeV ======== ------|-----------|------|--------------|------|--------------|---------------| | (e) | (e) | (e) | (p) | (p) | (e & p) | | | | | | | | SecNo | MMp +/-SIG| | Yt +/- SIG | | Yt +/- SIG | Phi Acollin. | | | | | | | & SIG of mean | | [MeV] | [cm] | [cm] | [cm] | [cm] | [Deg] | ======|===========|======|==============|======|==============|===============| | | | | | | | Sec#1 | 942 +/-12 |-0.96 |-2.10 +/-0.70 |-0.55 |-1.15 +/-0.67 |-2.37 +/-0.011 | | | | | | | | ------|-----------|------|--------------|------|--------------|---------------| | | | | | | | Sec#2 | 927 +/-11 |-0.39 |-0.48 +/-0.72 |-0.24 | 0.23 +/-0.58 |-1.00 +/-0.005 | | | | | | | | ------|-----------|------|--------------|------|--------------|---------------| | | | | | | | Sec#3 | 940 +/-17 |-0.82 |-0.11 +/-0.94 |-0.45 | 1.14 +/-0.59 |-0.83 +/-0.004 | | | | | | | | ------|-----------|------|--------------|------|--------------|---------------| | | | | | | | Sec#4 | 962 +/-15 |-1.69 |-0.32 +/-0.68 |-0.87 | 1.39 +/-0.63 | 0.01 +/-0.001 | | | | | | | | ------|-----------|------|--------------|------|--------------|---------------| | | | | | | | Sec#5 | 949 +/-12 |-1.76 |-0.71 +/-0.67 |-1.07 | 0.18 +/-0.68 | 0.74 +/-0.015 | | | | | | | | ------|-----------|------|--------------|------|--------------|---------------| | | | | | | | Sec#6 | 944 +/-14 |-1.84 |-1.73 +/-1.10 |-0.95 |-0.58 +/-0.69 | 1.49 +/-0.006 | | | | | | | | ------|-----------|------|--------------|------|--------------|---------------| Bogdan@jlab.org