ICARUS_PMT_calibration

PMT calibration scripts

Data is processed in a series of steps. First, preprocessing is performed using one of the Python scripts. Charge distributions are fit using one of the FitChargeDistribution scripts. Dark rate analysis can be performed separately using scripts in the dark_rate directory, while afterpulse analysis can be performed with scripts in the afterpulse directory.

Bash scripts used to read analysis outputs and write them to text in a format convenient for spreadsheet entry are located in the spreadsheet_compilation_scripts directory. Bash scripts for creating and running Condor analysis jobs are located in the condor_analysis_scripts directory.

Finally, old code is located in the old_code directory for archival purposes.

Preprocessing

To preprocess the raw data, PreprocessData.py is run. It assumes that raw waveform data was taken with a Tektronix MSO64 scope with the following settings:

• 4 channels,
• 20 us time window for each waveform
• each waveform is saved in a single file

The script does the following:

• a ROOT file is output with various analysis histograms and selected raw waveforms
  • the histograms include charge distributions
• the following information is printed:
  • the directory name with the waveforms
  • the number of waveforms for each PMT in this run
  • the number of pulses counted
  • the number of afterpulse candidates counted

To run the preprocessing analysis over an entire data set, a batch job can be submitted as follows:

1. a file named WaveformDirectories.txt must be created with the names of the waveform directories containing data. Only the directory names should be included, not the path to the directories. E.g. the directory at /media/disk_a/ICARUS/PMT/Data201906/B10_PMT_5_6_7_8_1440V_LedOff should be labeled as B10_PMT_5_6_7_8_1440V_LedOff.

   • this can be done, for example, using ls > WaveformDirectories.txt from within the directory of choice containing all of the data to be analyzed (e.g. /media/disk_a/ICARUS/PMT/Data201906)

2. change the data_path variable on line 3 of analyze_cl.sh to the proper directory (e.g. /media/disk_a/ICARUS/PMT/Data201906)

3. run the create_waveform_analysis_files.sh script

   • ensure that all the necessary files are present, including fix_led_names.sh
   • this will create individual folders for each group of data, to aid in parallelizing the analysis job

4. ensure that the input line in run_full_analysis_batch.job is correct. It should point to the directory you are working in. Also change the number after Queue at the end of the file to equal the number of directories created by create_waveform_analysis_files.sh

5. submit run_full_analysis_batch.job as a Condor job using condor_submit run_full_analysis_batch.job

For each directory containing data, the printed results of the preprocessing script will be written to a text file labeled as *output.txt, and the ROOT file will be output as usual.

Charge distribution fits

Standalone fits

FitChargeDistributions.C reads in the root file produced from PreprocessData.py, and fits charge distributions with Poisson function convoluted with a Gaussian function. The fit is for high charges, so a combined fit is performed using a single chi2 value and constant mu across three voltages.

• Processes four PMTs at a time
• Combined fitting code adapted from ROOT tutorials: combinedFit.C
• Output:
  • Root file
  • .pdf of the canvas contents
  • .txt of comma separated values including:
    • final parameters: voltage, channel number mu, mu error, q, q error, sigma, sigma error, amplitude, amplitude error, chi2, ndf, fit probability
    • initial parameters: channel id, start fit, end fit, rebin factor, mu, q, sigma, amplitude
• Uses parameter limits, which results in very poor errors
  • To improve the errors, save the parameters and rerun using FitChargeDistributions_InitParam.C
• The *LedOn*result.root files must be included in the same folder that this macro is stored in
• Do not have any *LedOff*result.root files in the folder
• Processes three root files in a batch (same four PMTs at three different voltages)
• Example function call in root: .x FitChargeDistributions.C("A10", 5, 6, 7, 8, 1400, 1430, 1460)
  • voltages as parameters must match the nominal values matching the *result.root files

FitChargeDistributions_LC_batch.C reads in the root file produced from PreprocessData.py, and fits charge distributions with a Poisson function convoluted with a Gaussian function. The fit is for low charges, so the model equation of a Poisson function convoluted with a Gaussian function is fit to each distribution individually (each PMT, each voltage); contrast to FitChargeDistributions.C

• Processes four PMTs at a time (contrast to FitChargeDistributions_LC_ind.C)
• Output:
  • Root file
  • .pdf of the canvas contents
  • .txt of comma separated values including:
    • final parameters: voltage, channel number mu, mu error, q, q error, sigma, sigma error, amplitude, amplitude error, chi2, ndf, fit probability
    • initial parameters: channel id, start fit, end fit, rebin factor, mu, q, sigma, amplitude
• Uses parameter limits, which results in very poor errors
  • To improve the errors, save the parameters and rerun using FitChargeDistributions_InitParam.C
• The *LedOn*result.root files must be included in the same folder that this macro is stored in
• Do not have any *LedOff*result.root files in the folder
• Processes three root files in a batch (same four PMTs at three different voltages)
• Example function call in root: .x FitChargeDistributions_LC_batch.C("A10", 5, 6, 7, 8, 1400, 1430, 1460)
  • Voltages as parameters must match the nominal values matching the *result.root files

FitChargeDistributions_LC_ind.C reads in the root file produced from PreprocessData.py, and fits charge distributions with a Poisson function convoluted with a Gaussian function. The fit is for low charges, so the model equation of a Poisson function convoluted with a Gaussian function is fit to each distribution individually (each PMT, each voltage); contrast to FitChargeDistributions.C

• Processes a single PMT at a time (contrast to FitChargeDistributions_LC_batch.C)
• Output:
  • Root file
  • .pdf of the canvas contents
  • .txt of comma separated values including:
    • final parameters: voltage, channel number mu, mu error, q, q error, sigma, sigma error, amplitude, amplitude error, chi2, ndf, fit probability
    • initial parameters: channel id, start fit, end fit, rebin factor, mu, q, sigma, amplitude
• Uses parameter limits, which results in very poor errors
  • To improve the errors, save the initial parameters and rerun using FitChargeDistributions_InitParam.C
• The *LedOn*result.root files must be included in the same folder that this macro is stored in
• Do not have any *LedOff*result.root files in the folder
• Processes three root files in a batch (same four PMTs at three different voltages)
• Example function call in root: .x FitChargeDistributions_LC_ind.C("A10", 5, 6, 7, 8, 1400, 1430, 1460, 0)
  • Last parameter is the index of the desired PMT out of the four, ie. 0, 1, 2, or 3
  • Voltages as parameters must match the nominal values matching the *result.root files

* NOTE: It is necessary when fitting using the above scripts to go into the code to adjust initial parameters if desired

Fits from .csv

FitChargeDistributions_InitParam.C is used to take in a .csv value of initial parameters and performs fits identical to those seen above, with the exception of the fact that parameter limits are no longer in place.

• Example table (to export as .csv) can be found here: https://docs.google.com/spreadsheets/d/19rzmrZNi8R2X_QgjkQUhmoAZPIGufCa9ip4rDvazvKE/edit?usp=sharing
• The macro utilizes a .csv file containing initial parameters for each PMT at three voltages in order to recreate the original fits.
  • Each PMT should have three rows dedicated to it
• The .csv file must contain two headers of the following form:
  • ,,,,Initial parameters,,,,,,,,Flags
  • ROOT File Name,Chimney,PMT channel,Voltage,Fit Begin,Fit End, Rebin Factor,Y Max,Mu (NPE),q (SPE),Sigma (SPE),Amplitude,Charge
  • There should not be any empty rows or columns in the .csv file
• The *LedOn*result.root files must be included in the same folder that this macro is stored in
• Do not have any *LedOff*result.root files in the folder
• Output (for each PMT):
  • Root file
  • .pdf of the canvas contents
  • .txt of comma separated values including:
    • final parameters: voltage, channel number mu, mu error, q, q error, sigma, sigma error, amplitude, amplitude error, chi2, ndf, fit probability
    • initial parameters: channel id, start fit, end fit, rebin factor, mu, q, sigma, amplitude
• Example function call in root: FitChargeDistributions_InitParam("data.csv")

Gain vs voltage analysis

The script GainVoltage.C performs analysis to determine the PMT gain as a function of voltage. It does the following:

• input: a text file containing the following 4 space-separated columns:

  1. PMT number
  2. PMT voltage
  3. gain
  4. gain error

  Each row is information for a single PMT at a single voltage. The order of the rows should not matter, nor should it matter if a PMT number is missing. Problems will arise if more than 6 rows contain information for a single PMT.

  The function input is the name of the text file, without .txt at the end. E.g. if the input file is A10.txt, then GainVoltage("A10") should be called.

• output:

  • PDF files with a gain-voltage fit for each PMT on both log-log and linear plots
  • ROOT file containing all gain-voltage fits
  • comma-separated text file containing fit parameters for each PMT, in increasing numerical PMT order

Dark rate analysis

Dark pulses are counted using the number of pulses counted in LED-off data. These pulses are counted in preprocessing. Currently, dark rates themselves are calculated externally, in a spreadsheet. The script dark_rate/HistogramDarkRate.C exists solely to produce histograms of the dark rates.

Afterpulse analysis

Most of the afterpulse probability determination is performed during preprocessing. The number of afterpulse candidate events is recorded then. In order to convert to a probability with dark rate subtraction, the Bash script afterpulse/calculate_afterpulse_probabilities.sh is used. The script assumes that three data sets are collected for each chimney --- the *.output files for the chimney must all be present in the same directory as the script. As an input, the script takes a chimney name, e.g. A10. It then searches for the output files with A10 at the front, and proceeds from there.

afterpulse/AnalyzeAfterpulse.C contains some preliminary code to do some basic afterpulse time-structure analysis.