Difference between revisions of "Tracking-1-21-2021"

From GlueXWiki
Jump to: navigation, search
(FDC minutes)
(FDC minutes)
Line 31: Line 31:
 
#* As reported in the PID meeting, Sarah (CMU) obtained a space charge correction matrix for dE/dx using pulse integral. This is described in [https://halldweb.jlab.org/DocDB/0046/004693/002/CDC_dE_by_dx_theta_dependence_and_corrections.pdf docdb 4693], the code is in github master and the matrix is in ccdb.  The best choice of whether to use integral or amplitude depends on the track momentum as well as the angle; the correction made a huge improvement to the dE/dx from integral. The dE/dx from integral is approximately 10% larger than that from amplitude, this arises somehow from the hit truncation.  The peak separation is best using pulse integral for medium to high momentum tracks at forward angles, and pulse amplitude for low momentum tracks and tracks close to 90 degrees. There are some comparison tables at the very end of the document for just a few angles and momenta.
 
#* As reported in the PID meeting, Sarah (CMU) obtained a space charge correction matrix for dE/dx using pulse integral. This is described in [https://halldweb.jlab.org/DocDB/0046/004693/002/CDC_dE_by_dx_theta_dependence_and_corrections.pdf docdb 4693], the code is in github master and the matrix is in ccdb.  The best choice of whether to use integral or amplitude depends on the track momentum as well as the angle; the correction made a huge improvement to the dE/dx from integral. The dE/dx from integral is approximately 10% larger than that from amplitude, this arises somehow from the hit truncation.  The peak separation is best using pulse integral for medium to high momentum tracks at forward angles, and pulse amplitude for low momentum tracks and tracks close to 90 degrees. There are some comparison tables at the very end of the document for just a few angles and momenta.
  
= FDC minutes =
 
 
#* FDC
 
#* FDC
 
#** FDC Wire Efficiency: Alex is working on a comparison of the (biased) FDC wire efficiency from Spring 2017 with MC. One quarter of the first FDC plane showed a high occupancy during the entire run, where an average wire efficiency of about 92% is measured. He tuned the <code>wire_mc_efficiency</code> table in CCDB to reproduce this deficiency in MC. This correction was implemented into the official database after the meeting. Remaining discrepancies either only affect a subset of the runs or only isolated wires and strips. These will be implemented once we have crosschecked the list with Lubomir.
 
#** FDC Wire Efficiency: Alex is working on a comparison of the (biased) FDC wire efficiency from Spring 2017 with MC. One quarter of the first FDC plane showed a high occupancy during the entire run, where an average wire efficiency of about 92% is measured. He tuned the <code>wire_mc_efficiency</code> table in CCDB to reproduce this deficiency in MC. This correction was implemented into the official database after the meeting. Remaining discrepancies either only affect a subset of the runs or only isolated wires and strips. These will be implemented once we have crosschecked the list with Lubomir.

Revision as of 10:21, 22 January 2021

  1. Instructions for Bluejeans meeting connection
  2. Meeting ID: 290664653
  3. To join via a Web Browser, go to the page [1] https://bluejeans.com/290664653.

Thu 1/21/2021 2pm

Agenda

  1. Tracking updates
  2. AOB
  3. Detector-related updates


Minutes

  1. FA125
    • In the problem run 72932, the CODA events PRESTART and GO were 9 minutes apart. A long gap between prestart and go used to cause problems with HOSS but Sergey thinks this problem was fixed some time ago. It was easy to find these using hd_dump <filename> -DDCODAControlEvent -s. The -s skips events without the requested data type. It would be easy to check the prestart-go interval for all runs if the 000.evio files are available.
    • The 1910 epics events that appear before the first physics event would take 9 minutes if written at 3 Hz, 28 minutes at 1 Hz. Sergey will check to see what the epics write frequency is.
    • When this problem was first discovered, Naomi checked all of the production runs for out-of-sync CDC fadcs and did not find any, other than 72932-72937 see logbook and follow-up entries. Lubomir checked FDC data. Naomi checked the CDC data using a plot of pulse amplitude for tracked hits from a plugin included in the monitoring launches, looking for channels with less than 1/4 of the average number of hits for straws in the same ring.
    • In general, the trigger times reported by Df125TriggerTime are identical for fadcs in slots 3-10 in each crate. They are also identical for fadcs in slots 13-17 in each crate. There is often a mismatch between the different crates and between the two groups in the same crate. David Abbott thought that this points to a SD problem, maybe firmware. Cody suggested that if the SD trigger is slightly out of spec, the fa125 might not see it.
    • Most runs don't seem to contain Df250TriggerTime, so we don't know if the fa250 issues uniform trigger times or not.
    • We would like to add software to the DAQ ROL's GO, to check that the fadcs are reporting identical trigger times (or if we don't fix the SD issue, that the fadcs in each half of each crate are reporting identical times). If they are not, then DAQ reset should fix it.
    • Sergey suggested adding monitoring code for rootspy that compares the trigger times similarly and prints an error message if it finds problems. He can also include something similar in DAQMON. This would work on a subset of events and would not require tracking.
    • Sergey plans to set up a DAQ with a pulser and the CDC crates, to look for startup problems. When this is set up, we can ask Cody to look at the SD signals with chipscope.
  2. CDC
    • As reported in the PID meeting, Sarah (CMU) obtained a space charge correction matrix for dE/dx using pulse integral. This is described in docdb 4693, the code is in github master and the matrix is in ccdb. The best choice of whether to use integral or amplitude depends on the track momentum as well as the angle; the correction made a huge improvement to the dE/dx from integral. The dE/dx from integral is approximately 10% larger than that from amplitude, this arises somehow from the hit truncation. The peak separation is best using pulse integral for medium to high momentum tracks at forward angles, and pulse amplitude for low momentum tracks and tracks close to 90 degrees. There are some comparison tables at the very end of the document for just a few angles and momenta.
    • FDC
      • FDC Wire Efficiency: Alex is working on a comparison of the (biased) FDC wire efficiency from Spring 2017 with MC. One quarter of the first FDC plane showed a high occupancy during the entire run, where an average wire efficiency of about 92% is measured. He tuned the wire_mc_efficiency table in CCDB to reproduce this deficiency in MC. This correction was implemented into the official database after the meeting. Remaining discrepancies either only affect a subset of the runs or only isolated wires and strips. These will be implemented once we have crosschecked the list with Lubomir.
      • FDC Simulation of Lorentz distortions: If we correlate the distance of the track to the wire with the relative position of the cluster to the track along the wire, we are able to measure the Lorentz-effect on the drifting charged in the magnetic field. The slope of this shift is a measure of the magnetic field. In MC, we can also observe this effect, but the magnitude is 25% smaller. The Lorentz effect is parametrized by the CCDB constants lorentz_deflection_parms, which are used both for simulation and for the correction of the effect in the tracking algorithm. After the meeting, we discovered that the leading parameter in the mc variation is indeed ~25% smaller. We plan to update it after verifying that MC agrees with the measured effect. The net effect on the tracking efficiency is expected to be small.

TRD minutes

    • TRD

We can merge the minutes sections later, I separated them so that we can edit each section without tripping each other up.