2005 Cross Talk Analysis




Summary of 2005 Cross Talk Analysis:

The cross-talk analysis for AMANDA strings 5-10 was done using T0 data from the 04-05 South Pole season taken in November and December of 2004 by Elisa Bernardini and Mike Walter. The data is currently maintained on pub.ifh.de. Please refer to the T0 DESY website for more information on the T0 callibration. Click here for a list of the runs and associated data files used from the T0 callibration for this cross-talk analysis.The cross-talk constants were obtained by fitting the ADC-TOT curves for each OM using the PAW kumac contained in xt_2000 as part of the xt-filt software package developed and maintained by Ignacio Taboada. Some of the OM's were fit using down-going muon data in place of T0 data. An explanation of this can be found below in the Problems Encountered section. The Correlation Map was obtained from all available T0 data by using Ignacio's xt_anal program to check for correlated hits in the T0 data.


Data Processing Specifics (ADC-TOT Fits):

The raw data files were downloaded from the IceCube server at University of Wisconsin (pub.icecube.wisc.edu) and processed on the AMANDA server at Berkeley. Click here for a list of all the data files used in the cross-talk analysis. The following were the steps taken in reading and filtering the raw data:

  1. convert raw data from binary format to f2k format using the AMANDA raw data reader (v2.4) with 2002 detector settings-
  2. amanda:~/data/ amorey> reader.2.4 -i file.bin.gz -f file.f2k --run-period=2002

  3. calibrate the f2k data file for detector geometry using amacalib (Siegmund softare package)-
  4. amanda:~/data/ amorey> amacalib -L -P -T -D ~/xt-filt/omdb-00-20010926.amacalib.txt file.f2k.gz -o calib.file.f2k.gz

  5. remove fake OM hits (data from OM's numbered 9000-9009 or 10000-10999) using grep-
  6. amanda:~/data/ amorey> zcat calib.file.f2k.gz | grep -v "HT 9[0-9][0-9][0-9]" | grep -v "HT 1[0-9][0-9][0-9][0-9]" | gzip -c > clean.calib.file.f2k.gz

  7. convert f2k file format to hbook format using munt (Siegmund software package)-
  8. amanda:~/data/ amorey> munt -K -O clean.calib.file.hbook clean.calib.file.f2k.gz >& /dev/null


Data Processing Specifics (Correlation Map):

The raw data files used for the Correlation Map were the same as the T0 data files used for the ADC-TOT fits in addition to T0 data for OM's 92, 99, 119, 126, 130, 175, 187, 196, 198, 222. I was unable to read these data files with PAW for the ADC-TOT fits but I was able to use them to obtain the correlation map. The following were the steps taken in obtaining cross talk pairs from the raw T0 data files:

  1. convert raw data from binary format to f2k format using the AMANDA raw data reader (v2.4) with 2002 detector settings-
  2. amanda:~/data/ amorey> reader.2.4 -i file.bin.gz -f file.f2k --run-period=2002

  3. calibrate the f2k data file for detector geometry using amacalib (Siegmund softare package)-
  4. amanda:~/data/ amorey> amacalib -L -P -T -D ~/xt-filt/omdb-00-20010926.amacalib.txt file.f2k.gz -o calib.file.f2k.gz

  5. check for cross talk candidates for OM# being flashed in T0 run using xt_anal (xt-filt software package)-

    amanda:~/data/ amorey> xt_anal calib.file.f2k.gz -v -o file.corr.hbk -s OM#


ADC-TOT fits:

The ADC-TOT curves for the T0 data were fit using a PAW kumac created by Ignacio Taboada. Each graphic shows the points chosen to represent the ADC-TOT edge, the best-fit curve, the standard TOT cut-off, and the actual ADC-TOT curve which will be used in cross-talk cleaning. Below is an example of an ADC-TOT curve fit:

Below is a complete list of the fit status for each OM along with graphic files showing the best-fit curves. Most of the OM's were fit without any issues using the T0 date while several of the OM's were fit using muon data. Please see the Problems Encountered section for a more detailed description of the issues encountered in fitting the ADC-TOT curves.


2005 Cross Talk Constants:

The cross-talk constants are finished. Please refer to the Problems Encountered section for a discussion of the problems encountered and the solutions provided. Click on the link below for a text file containing the constants in their standard format.


2005 Correlation Map:

A preliminary version of the correlation map is ready. Click on the link below for a text file containing a list of all cross-talk pairs with the associated ΔLE_lo and ΔLE_hi for each pair. The only cuts used in obtaining the cross-talk pairs from all possible candidates were that the number of correlated hits be greater than 10 and that 90% of all correlated hits have TOT>10. I have not yet finished obtaining the ADC thresholds and widths for the cross talk pairs. In the mean time I have posted a standard turn-on threshold (ADC at 50% cross-talk probability) of 2000 mV and a width of 50 mV. Please refer back to this web page, as the constants will be available shortly.

The following are scatter plots of cross-talk pairs by string. The red diagonal line in each plot is the line of symmetry and those pairs on the outside of the black diagonals are spaced greater than 15 om's apart (top-to-bottom cross talk):
Problems Encountered:

  1. Dead OM's (click here for a complete list of OM status as of the beginning of 2005):

  2. The OM's listed above show up as "dead" in the ADC-TOT Curve Fits and with 0's for their Cross-Talk Constants.


  3. Live OM's fit with down-going muon data:

  4. In place of T0 data, I used muon data from March 12-14, 2005 (run numbers 9185, 9186, and 9187) to fit the ADC-TOT curves for the above OM's.

    Explanations:
  5. OM's with awkward fits:

  6. These OM's were unusually difficult to fit using the PAW kumac and the thoeretical ADC-TOT curves. The final cross-talk constants for these OM's were chosen to give as close an approximation to the curve as possible while not eliminating real data. Please refer to to the figures for the OM's to verify that the ADC-TOT cuts will not eliminate real data.


  7. OM's with "double-edged" data:

  8. The muon data for OM 99 gave two distinguishable ADC-TOT edges and was fit using the outer edge. This was chosen as the safer edge so as to avoid eliminating real data. Please refer to the figure for OM 99 to verify that the fit is acceptable.


  9. The T0 data was read using the AMANDA raw data reader with 2002 settings. As of 4/27/05, the 2004 detector settings for the reader were not available.

  10. Initially, I was finding that many of the OM's had "insufficient T0 data". Elisa pointed out that she was able to produce ADC-TOT plots with a sufficient number of hits from the same data. Upon closer inspection, we found that the problem was in the leading edge distribution for the OM's and our time-window optimization algorithm. After-pulsing in some of the OM's confused our algorithm and we were viewing the hits in between the two pulses rather than at the peak of the main pulse. This problem was corrected and those OM's found to have insufficient data previously, were re-fit.

  11. The hit structure for many of the OMs appears to be shifted down from previous years. Here is an example of ADC-TOT data taken from the last T0 run as compared to data from 2000:
  12. T0 2005 2000 (courtesy of Henrike)

    It can be seen that hits which were below 100 mV in the year 2000 were not recorded in the 2004-2005 T0 run. The data seems to be shifted downwards. This could be a problem if good single photoelectron hits are being ignored. The change in hit structure seems to be due to the change in the high-voltage settings for the OMs in question.



Last updated Tuesday, August 8, 2005.