Offline Monitoring Data Validation

From GlueXWiki
Jump to: navigation, search

This page contains the procedure for checking if production runs are of good quality and can be used for physics analysis.

Run Periods


For each production run, do the following:

  • Go to the Offline Run Browser page.
  • Follow the steps outlined in the checklist below.
  • Workers should check each plot for their assigned subsystem and leave notes in the corresponding spreadsheet if any significant deviations are seen
  • On the spreadsheet, enter "Y" in the "Overall Quality" field if all monitoring histograms are acceptable, otherwise enter "N"
  • We will iterate this procedure until the process converges

Expert Actions

  • Certify that each subsystem is okay
  • Set run status in RCDB based on monitoring results
    • (script provided)

Run Statuses

  • -1 - unchecked
  • 0 - rejected (not physics-quality)
  • 1 - approved
  • 2 - approved long/"mode 8" data
  • 3 - calibration / systematic studies


Example Monitoring Spreadsheet

Reference run for 2017-01: 30780

Reference run for 2018-01: 40933/4

Reference run for 2018-08: 51388

Reference run for 2019-11: 71463 (150 nA), 71469 (250 nA), 71724 (350 nA)

Reference run for 2023-01: 120888

Expert list

Experts should update the table as tasks are completed

Instructions status by subdetector
Subdetector Plots Instructions Expert(s)
BCAL Good Good Mark Dalton, Zisis Papandreou
CDC Good Good Naomi Jarvis
DIRC  ?  ? Justin Stevens
FCAL  ?  ? Mark Dalton, Malte Albrecht, Igal Jaegle
FDC Good Good Lubomir Pentchev
PS Good Good Alex Somov, Olga Cortes
SC Good Good Beni Zihlmann
TAGH Good Good Alex Somov, Bo Yu
TAGM Good Good Richard Jones, Ellie Prather
TOF Good Good Paul Eugenio, Beni Zihlmann
RF Good Good Sean Dobbs, Beni Zihlmann
Timing Good Good Sean Dobbs
Analysis Good Good Alex Austregesilo

General Notes

  • Diamond and amorphous (AMO) runs have different beam energy spectra, which leads to differences in reaction yield distributions which depend on the kinematics of the produced particles.


  • Check Occupancy - Reference: [ link ]
  • Check Hit Efficiency - Reference: [ link ]

BCAL Reference Plots

BCAL Notes

The BCAL is used to measure the energy and time of showers.

  1. Occupancy: This should be approximately flat. There can be hot channels when the baseline drifts.
  2. Hit Efficiency: This should be approximately flat. If there are features we should understand why.


  • Check Occupancy - Reference: [ link ]
  • Check Time-to-distance - Reference: [ link ]
  • Check dE/dx - Reference: [ link ]
  • Check Efficiency- Reference: [ link ]

CDC Reference Plots

CDC Notes

CDC Occupancy: There should be a uniform decrease in intensity from the center of the detector outward. Random white cells scattered throughout occur when not enough data were collected, eg empty target runs, trigger tests or no beam. Several contiguous white, dark blue or bright yellow cells which don't match the neighboring cells are a problem.
Time-to-distance: 𝛿, the change in length of the LOCA caused by the straw deformation, is plotted against the measured drift time, t drift . The color scale indicates the distance of closest approach between the track and the wire, obtained from the tracking software. The red lines are contours of the time-to-distance function for constant drift distances from 1.5 mm to 8 mm, in steps of 0.5 mm. They should lie over the top of the dark blue contour lines separating the colour blocks. For the plot of residuals vs drift time, the mean should be less than 15um and the sigma should be less than 150um.
dE/dx: At 1.5GeV/c the fitted peak mean should be within 1% of 2.02 keV/cm.
Efficiency: The efficiency should be 0.98 or higher at 0cm DOCA, gradually fall to 0.97 at approximately 0.5mm and then more steeply through 0.9 at approximately 0.64cm.


  • Check N Occupancy - Reference: [ link ]
  • Check S Occupancy - Reference: [ link ]

DIRC Reference Plots

DIRC Notes



  • Check Occupancy - Reference: [ link ]

FCAL Reference Plots

FCAL Notes

Is used for neutral particle detection and pion identification.

Check Occupancy:
For monitoring purposes, the occupancy of the detector should be checked for every run once - so this is only needed for the first ever monitoring launch per run period. The goal is to find any blocks that do not deliver a signal for each run, these must be made dead channels in the Monte Carlo simulation for that specific run. Watch out for individual blocks as well as groups of 16 channels in a 4x4 orientation, which indicates a faulty fADC.


  • Check Package 1 Occupancy - Reference: [ link ]
  • Check Package 2 Occupancy - Reference: [ link ]
  • Check Package 3 Occupancy - Reference: [ link ]
  • Check Package 4 Occupancy - Reference: [ link ]

FDC Reference Plots

FDC Notes

There are two HV sectors, in Package 2 cell 6 (28 wires) and Package 3 cell 4 (20 wires), that are always OFF and seen in the occupancy plots as empty sectors. There are also strips with lower or no efficiency that are always there, mostly in Package 3 and 4 (see the reference plots), which also normal. What is not normal are groups of wires (of the order of 8 to 24 wires) that are noisy. They will show as brighter stripes in the occupancy. The problem is that they may lock the F1TDCs. This happened several times in the past years. In general, look for groups of channels that are overactive or have lower efficiency.

The reference plots show pseudohits, generated from the track reconstruction. If you find an abnormality, it would be helpful to check 2 more histograms to find the underlying cause - 'FDC Hit Occupancy' will show if any channels are missing, and 'HLDT Drift Chamber Timing' (bottom right plot) will show TDC time-shifts.


  • Check Occupancy - Reference: [ link ]
  • Check Timing Alignment - Reference: [ link ]
  • Check PS Pair Energy - Reference: [ link ]

PS Reference Plots

PS Notes

PS Occupancy: PS Occupancy (bottom) should be fairly flat with a couple bad channels. PSC Occupancy (top) should have similar rates in TDC and ADC, with the same shape as the reference histogram.

PS Timing: All plots should be centered at zero. The right column reflect the tagger energy, the bottom right is empty (should be updated?).

PS Pair Energy: Should have similar triangle-like shape as the reference.


  • Check Occupancy - Reference: [ link ]
  • Check Recon. SC 1 - Reference: [ link ]
  • Check Recon. SC 2 - Reference: [ link ]
  • Check Recon. SC Matching - Reference: [ link ]

SC Reference Plots

SC Notes

  • Occupancy: check for "gaps" indicative of HV off/trip
  • Recon SC1: top middle, visible dEdx proton "band" below 1 GeV/c
  • Recon SC2: timing difference with RF all peak at zero
  • Recon SC3: bottom right most >90% between 60mc and 95cm, bottom middle all paddles above 90%


  • Check Tagger Occupancy - Reference: [ link ]
  • Check TAGH Hits 2 - Reference: [ link ]

TAGH Reference Plots

TAGH Notes

Tagger occupancy: TAGM - Generally the fADC and TDC occupancies should be similar and mostly flat, with maybe a small increase in rates with column number. There can be a some steps in the TDC occupancy. TAGH - expect the choppy pattern in the reference image, which reflects the varying size of the different counters, and a steep increase at large counter number.
TAGH Hits 2: This plot is complicated - the main thing to look for is the time(TDC)-time(ADC) vs. channel plot to be centered around zero. Keep an eye out for any extra or unusual dead channels.


  • Check Timing ADC-RF - Reference: [ link ]
  • Check Timing T-ADC - Reference: [ link ]

TAGM Reference Plots

TAGM Notes

Generally both distributions should be centered near zero. There is some variation in intensity due to the shape of the photon beam energy dependence (coherent peak) and the inefficiency of some of the channels.


  • Check Occupancy - Reference: [ link ]
  • Check TOF Matching 1 - Reference: [ link ]
  • Check TOF Matching 2 - Reference: [ link ]

TOF Reference Plots

TOF Notes

  • Occupancy: Check all counters are present, "no gaps" this is indicative of HV off/trip
  • Matching 1: right most column, top and bottom the intensity should peak close to zero and should form a ridge along zero (in y).
  • Matching 2: bottom left should look symmetrically "round/circle with hole", bottom left between 20cm and 65cm should be above 80%


  • Check timing offsets - Reference: [ link ]
    • Should be centered around zero

RF Reference Plots


  • Check HLDT Calorimeter Timing - Reference: [ link ]
  • Check HLDT Drift Chamber Timing - Reference: [ link ]
  • Check HLDT PID System Timing - Reference: [ link ]
  • Check HLDT Tagger Timing - Reference: [ link ]
  • Check HLDT Tagger/RF Align 2 - Reference: [ link ]
  • Check HLDT Tagger/SC Align - Reference: [ link ]
  • Check HLDT Track-Matched Timing - Reference: [ link ]

Timing Reference Plots

Timing Notes

  • Calorimeter Timing - The right two plots aren't aligned at zero because not all corrections are currently applied. If there is a 32 ns shift in part of this data, please note this.
  • Drift Chamber Timing - In each case, the main peaks should line up at zero, but often have other structures. Ignore the first few bins of the lower left plot (they mostly say something about the noise in the detector). There can be 32 ns shifts in the lower right plot. The yellow vertical bar on the bottom right plot is not typical, but is an example of a noisy chamber. These should be noted, but marked as okay.
  • PID System Timing - Look for peaks aligned around zero.
  • Tagger Timing - The signal to background levels of the left two plots depend on the electron beam current. Ignore any peaking at the left edge of the distribution.
  • Tagger/RF Timing - Look for the nice "picket fences" on the right two plots, and that in the bottom left plot each channel peaks at zero.
  • Tagger/SC Timing - Should be similar to Tagger/RF Timing but with larger resolution.
  • Track Matched Timing - Some overlap here with the tracking timing. The new plots should be centered at zero.


  • Tracking 1 - [ link ]
  • Tracking 3 - [ link ]
  • Check BCAL pi0 - Reference: [ link ]
  • Check BCAL/FCAL pi0 - Reference: [ link ]
  • Check p+2pi - Reference: [ link ]
  • Check p+3pi - Reference: [ link ]
  • Check p+pi0g - Reference: [ link ]

Analysis Reference Plots

Analysis Notes

Generally in these plots, there will be a difference between diamond and amorphous radiator running. Should probably add some references for non-diamond plots.

  • Tracking 1 - There should be some mild dependence on beam current and radiator. Note the spikes in the upper right plot are because we have 4 hypotheses fit to a track by default. The lower left plot does have a peak at zero.
  • Tracking 3 - All four plot should have the pion band around 2keV/cm. Only the top left one should have an additional banana-shaped band for the protons.
  • Check BCAL pi0 - The fitted peak should near at the correct pi0 mass of 135 MeV.
  • Check BCAL/FCAL pi0 - The fitted peak should be lower than the correct pi0 mass, I think because the wrong vertex is used.
  • Check p+2pi - The top middle plot should have a sin(2phi) shape for diamond runs. Note that the yields in the top right plot vary from run to run on the order of 10-20%.
  • Check p+3pi -Note that the yields in the top right plot vary from run to run on the order of 10-20%.
  • Check p+pi0g - Note that the yields in the top right plot vary from run to run on the order of 10-20%.
    • Note that these yields are sensitive to the tagger range used! This changes for different beam current settings.