GlueX Start Counter Meeting, July 23, 2020

From GlueXWiki
Jump to: navigation, search

GlueX Start Counter Meeting
Thursday, July 23, 2020
10:00 am EDT
BlueJeans: 556 286 544


  1. Announcements
  2. Review minutes from the last meeting
  3. Calibration and Efficiency
  4. Dark Rate Studies
  5. Action Item Review


Present: Rupesh Dotel, Mark Ito (chair), Simon Taylor, Beni Zihlmann

There is a recording of this meeting on the BlueJeans site. Use your JLab credentials to gain access.


The start of the run has been delayed. All shifts are canceled until August 3rd.

Review of minutes from the last meeting

We went over the minutes from July 9.

Run Readiness

Beni reported the start counter is ready to go, all channels look good electronically. The only thing left to check is that the data appears properly in the events output by the DAQ. The start counter has been moved to Ready in the Hot Check-Out list.

Pulse-Height Data quality

Rupesh performed fits for individual channels of the start counter to the distribution of peak amplitude of the ADC hits. He fit to a Landau distribution in a limited region of the distribution to avoid noise hits at threshold. Data came from a single run. He did this for several runs. Here are the fits from channel 30:

Sc fit 30274.png Sc fit 31057.png
Sc fit 40856.png Sc fit 42559.png
Sc fit 71350.png Sc fit 71839.png

From the fit parameters, he formed the ratio of the "width" to the most-probable value of the Landau distribution, for each channel and for each of these runs, and plotted them:

Sc ratios.png

There is no clear trend. We asked if he could plot the level found in each run, averaged in some way over all channels, as a function of beam intensity and as a function of time so see if any trends emerge.

Copying input files to the local disk on the farm

Mark sent a question to Ying Chen on the subject. Here is the email exchange:


What is the current idea for how to analyzing data from cache in a farm job, in particular for raw data? Should one copy the file from cache to the local disk and analyze it from there, or should one analyze the data directly from cache? My memory is that the advice on this has changed in years past and I don't know which is the current best practice.

-- Mark


I think this depends on how the job access the file. If it just read once, it should be good to read from /cache directly.

If there are many small read, we suggest user to copy the file to local disk.</div>

By default, Auger will not copy the files to local disk when the input files from /mss (only make a link), and files from other location (/volatile, /work, /home etc.) will be copied to local disk. The reason behind this design is that the files stored on /cache are larger file, on /home, /work and /volatile are small files. But in later 2018, I add a new attribution (copyOption = link or copy) so user can specified copy to link the file.

But I don't know whether Swif implements this option yet. If not, user still can use the old way to copy the file to local disk.


Action Item Review

  1. Document calibration procedure (Rupesh)
  2. Reproduce Beni's dark current results. (Tolga)
  3. Revisit Yi's study of SiPM annealing. (Mark, Beni)
  4. Look up type of SiPMs used in start counter. (Mark)