glast lat project instrument analysis workshop 5 – 05/08/29 f. piron & e. nuss (in2p3/lpta –...

GLAST LAT Project Instrument Analysis Workshop 5 – 05/08/29

F. Piron & E. Nuss (IN2P3/LPTA – Montpellier) 1

Comprehensive review of Comprehensive review of CAL calibrationsCAL calibrationsGamma-ray Large Area Gamma-ray Large Area

Space TelescopeSpace Telescope

• Trending CAL performance parameters as a function of LAT assembly phase:– from Comprehensive Performance

Tests– from calibration files (calibGenCal)

• Mapping CAL crystals and energy response

• Conclusions



Trending CAL performance parameters : an introduction 1/5Trending CAL performance parameters : an introduction 1/5

• Let’s start with an example: pedestals

• Define parameters– pedestal positions– pedestal widths

• Define phases– pre-ship (“Pshp”)– post-ship (“SLAC”)– 8 towers in GRID (“8T”)

• Trend parameters in 3 steps– Raw plots: plot raw values for each channel– History plots: plot parameters variation w.r.t. first phase

(“Reference=Pshp”)– Summary plots: to have a first quick look

• Store plots in 1 pdf file per CAL module (147 pages for calibGenCal trending, 133 for CPTs !!) and scan it!




1 face

distribution

raw plots

4 energy ranges

48 channels

Pedestal widthsPedestal positions




1 channel1 energy

range

history plots

3 phases

Variation (phase 0 ALWAYS 0)

Pedestal positions Pedestal widths

Change of TPS before inserting in the GRID




1 face

history plots

4 energy ranges

Mean+-RMS,

MIN and MAX

48 channels

RMS,

0.5(MAX – MIN)

Pedestal widthsPedestal positions




summary plots

RMS and 0.5(MAX – MIN), summed over all last histograms

Sometimes expressed in percentage

Change of TPS before inserting in the GRID



Trending CAL performance parameters from CPT’s 1/15Trending CAL performance parameters from CPT’s 1/15

• What do we trend ?– Compute pedestals: we trend position and width– Check optical response (with muons):

• a test for changes in the PDA (photo-diode assembly) optical bond quality is made from the ratio of LE or HE diode signals

• we trend LE+/LE-, LE+/HE+ and LE-/HE-– Determine and trend optimal charge injection time delay (TACK

DELAY) under timed readout– Calibrate electronic gains (with CI):

• we trend the lowest and highest relative gains (w.r.t. nominal gains)

• these gains are chosen because any drift from the nominal value would be most greatly amplified

– Determine front-end integral non-linearity and noise (with CI): we trend noise, non-linearity (deviation from linear fit in %) and DAC2ADC slope

– Characterize FLE, FHE, LAC and ULD DAC settings (with CI): we trend the coarse range DAC2ADC slopes

• Which phases ? Pre-ship, Post-ship, 8T




T00-FM104




T04-FM105




T01-FM103




T05-FM102




T08-FM107




T09-FM106




T12-FM110




T13-FM108




T12-FM110

• Lack of statistics ?




T00-FM104 T04-FM105




T01-FM103 T05-FM102




T08-FM107 T09-FM106




T12-FM110 T13-FM108




Check the fit for these guys…



Trending CAL performance parameters from calibration files 1/16Trending CAL performance parameters from calibration files 1/16

• What do we trend ?– Compute pedestals: we trend position and width

– ADC response: • fit DAC to (pedestal subtracted) ADC in a region away from

non-linearities• we trend xing value and DAC2ADC slope

– Characterize asymmetry functions for small (S) and large (L) diodes:• Compute left/right ratios (L/L, L/S, S/L or S/S) as a function of

log segment (except 2 segments at the log ends)• fit a linear function (though not well suitable...)• we trend xing value and slope

– Energy calibration: we trend the small and large diode MeV per DAC constant

• Which phases ? Pre-ship, 6T, 8T




T00-FM104




T01-FM103




T05-FM102




T08-FM107




T09-FM106




T12-FM110




T13-FM108




T00-FM104

T01-FM103

Energy gains are larger

(and stable) in the GRID




T00-FM104




T01-FM103




T05-FM102




T08-FM107




T09-FM106




T12-FM110




T13-FM108



Mapping CAL crystals and energy responseMapping CAL crystals and energy response

• Use TKR tacks (Tkr1XYZDir) and CAL tracks (CalXYZDir)

• Select events (% is given for 2 towers)– CalZDir<1 (90%)– TkrNumTracks==1 (80%)– >6 hits above 2 MeV in at least one tower (25%)

• Data set:– MC: 2 tower surface muons (22%)

– Data (Merit and Svac tuples):• 2T : B/2 + B/10 + B/13 = 20h20’ (18%)• 4T : B/10 + B/13 = 2h00’ (22%)• 6T : B/2 + B/10 + B/13 = 20h00’ (24%)



Comparing TKR and CAL extrapolated tracksComparing TKR and CAL extrapolated tracks

start of track

end of track

multiple scattering ??

From Benoit’s note (4GeV)

RMS of difference in

lateral position

RMS of difference in longitudinal position

E < 2 GeV E > 3.5 GeV

0.8 mm

1.4 mm



Comparing TKR extrapolated track with MC truthComparing TKR extrapolated track with MC truth

E < 2 GeV

E > 3.5 GeV0 mm

2 mm

0.3 mm

0.9 mm

multiple scattering effect is dominant

RMS of difference in lateral position


E < 2 GeV

E > 3.5 GeV



Comparing CAL extrapolated track with MC truthComparing CAL extrapolated track with MC truth

RMS of difference in lateral position


14 mm

0 mm

TKR-CAL discrepancy is not due to multiple scattering in CAL

E < 2 GeV E > 3.5 GeV

E > 3.5 GeVE < 2 GeV



Mapping CAL crystals using TKR extrapolated tracksMapping CAL crystals using TKR extrapolated tracks

Divide each crystal in 9 (3mm) * 54 (6mm) bins

Map of hits per bin Map of mean energy per bin



Mapping CAL crystals using TKR extrapolated tracksMapping CAL crystals using TKR extrapolated tracks

Map of hits per bin

50

140

0 0 MeV

20 MeV



Crystal longitudinal energy response profile (tower0, Crystal longitudinal energy response profile (tower0, layer0)layer0)

2 towers 6 towers

-10 %

+10 %



Distribution of MPV’s – 2 and 6 towersDistribution of MPV’s – 2 and 6 towers

2 towers:

<MPV> = 11.25 +- 0.007 MeV

6 towers:

<MPV> = 11.58 +- 0.004 MeV



Distribution of MPV’s – 4 towersDistribution of MPV’s – 4 towers

4 towers: old calibration

2 peaks at 10.8 and 11.2 MeV

4 towers: new calibration

<MPV> = 11.41 +- 0.05 MeV



• Trending and crystal mapping show calibration is stable– variations in pedestals and MeV/DAC related to change of

TEM power supply– definition of some parameters has to be improved (e.g.

asymmetry slopes)– Need for a better way to extract csv and calibration files from

database to extend the trending to other LAT assembly levels

• To do:– redo 2 tower calibration using muons from B10 runs and CI

from 8T (4T calibration already re-done this way)– reprocess 2T data using this new calibration (4T already

available)– calibrate flight gains using muon gain calibration + relative

gains

ConclusionsConclusions

glast lat project instrument analysis workshop 5 – 05/08/29 f. piron & e. nuss (in2p3/lpta –...

Documents

nuss in2p3lpta montpellier

t trend parameters

cal module

grid slide

calibgencal trending

t slide

plot parameters variation

channel history plots