glast lat project lat instrument analysis meeting – may 26, 2006 hiro tajima, tkr efficiency...
Post on 21-Dec-2015
218 views
TRANSCRIPT
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 1
GLAST Large Area Telescope: GLAST Large Area Telescope:
TKR Efficiency Trending
Hiro Tajima (SLAC)
TKR
Gamma-ray Large Gamma-ray Large Area Space Area Space TelescopeTelescope
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 2
Efficiency Trending Result @ IA WorkshopEfficiency Trending Result @ IA Workshop
• Consistent downward trend observed.
– Inconsistent with stable bad strips.
– Probably due to LAT configuration change.
• Efficiency values depend on track quality and other factors.
• Further improvement on track selections required to make it more stable.
Efficiency Trend
-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
2TWR 4TWR 6TWR 8TWR 16TWR
Test Phase
Efficiency change (%)
TkrFMA
TkrFMB
TkrFM1
TkrFM2
TkrFM3
TkrFM4
TkrFM5
TkrFM6
TkrFM7
TkrFM9
TkrFM10
TkrFM11
TkrFM12
TkrFM13
TkrFM14
TkrFM15
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 3
Improvement on Efficiency CalculationImprovement on Efficiency Calculation
• Use only towers with 6 or more hit layers.
– Much improved. Still more scatter than the statistical error.• FMA: lower efficiencies in 16-tower and first LICOS muon data. • FM4: lower efficiencies in PSR muon data.
– Systematically lower efficiency at NRL.• So is trigger rate. (~500 Hz @SLAC ⇒ ~420 Hz @NRL)
FM# Hand-off 16-tower difference LICOS difference PSR difference PER(NRL) differenceA 98.59% 98.54% -0.05% 98.53% -0.06% 98.58% -0.01% 98.55% -0.04%B 99.84% 99.82% -0.02% 99.82% -0.02% 99.83% -0.01% 99.82% -0.02%1 99.74% 99.72% -0.02% 99.72% -0.02% 99.73% -0.01% 99.71% -0.03%2 99.74% 99.74% 0.00% 99.73% -0.01% 99.75% 0.01% 99.74% 0.00%3 99.67% 99.69% 0.01% 99.69% 0.02% 99.70% 0.03% 99.69% 0.02%4 99.84% 99.81% -0.03% 99.80% -0.04% 99.76% -0.08% 99.78% -0.06%5 99.87% 99.92% 0.05% 99.92% 0.05% 99.92% 0.05% 99.90% 0.03%6 99.81% 99.79% -0.01% 99.80% -0.01% 99.81% 0.00% 99.78% -0.03%7 99.87% 99.85% -0.02% 99.83% -0.04% 99.86% -0.01% 99.83% -0.04%9 99.90% 99.90% -0.01% 99.89% -0.01% 99.90% 0.00% 99.84% -0.06%
10 99.85% 99.85% 0.00% 99.85% 0.00% 99.86% 0.01% 99.83% -0.02%11 99.88% 99.88% 0.00% 99.87% 0.00% 99.87% -0.01% 99.81% -0.06%12 99.93% 99.91% -0.02% 99.91% -0.02% 99.91% -0.02% 99.85% -0.08%13 99.90% 99.90% -0.01% 99.88% -0.02% 99.90% 0.00% 99.87% -0.04%14 99.88% 99.86% -0.01% 99.86% -0.01% 99.87% -0.01% 99.85% -0.02%15 99.90% 99.89% -0.01% 99.88% -0.01% 99.89% 0.00% 99.83% -0.06%
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 4
FM4 (Bay 13) Layer EfficienciesFM4 (Bay 13) Layer Efficiencies
• Layer X13 (plane# 26) shows significant efficiency drop.
+ LICOS runs+ PSR runs
Layer ID
Eff
icie
ncy
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 5
FM4 X13 Strip-Level EfficienciesFM4 X13 Strip-Level Efficiencies
• This is not exact strip efficiency due to accuracy of track extrapolation.– It gives crude association of efficiency with strip#.– Efficiency of strips 1024-1152 degraded significantly.– This region had lower efficiencies in previous runs.
Strip ID
Eff
icie
ncy
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 6
SSD Wafer Dependence?SSD Wafer Dependence?
• Occupancy does not depend on SSD wafer in the ladder.
• Efficiency loss is well correlated with GTFE boundaries.
Strip ID
Occ
up
ancy
(ar
bit
rary
un
it)
Strip ID
GTFE boundaries
Ladders
GTFE
Wafer 0
Wafer 3
Wafer 2
Wafer 1
– Wafer 0– Wafer 1– Wafer 2– Wafer 3
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 7
• New hot strip around 1141, 1143 (close to edge of the SSD)!– 1142 was masked in previous runs.– Hot strips is due to large leak current from junction break down.
• Mechanism for efficiency loss in two GTFEs is not understood.• New hot strip schema files released to mask these hot strips.
– Mainly to reduce trigger occupancy.– It may not fix the efficiency loss since we had some efficiency
loss even when these strips were quite.
Root Cause of Efficiency LossRoot Cause of Efficiency Loss
LICOS runs PSR runs
Strip ID
Occ
up
ancy
(ar
bit
rary
un
it)
Strip ID
1142
1141, 1143
Sugizaki
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 8
FMA (Bay 0) Layer EfficienciesFMA (Bay 0) Layer Efficiencies
• Slight efficiency loss in Y0, Y3, Y4.
+ LICOS runs+ PSR runs
Layer ID
Eff
icie
ncy
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 9
FMA Y3, Y4 Strip-Level EfficienciesFMA Y3, Y4 Strip-Level Efficiencies
• No localized significant efficiency loss observed.– Some intermittent strips in green circles could be a cause.
disconnected strips
ladder disconnected strips
ladder disconnected strips and healthy strips are mixed and averaged out.
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 10
Automated Layer Efficiency TrendingAutomated Layer Efficiency Trending
• Tower efficiency is not very sensitive to local efficiency loss.– Useful for overall efficiency trending.
• Layer efficiency is more sensitive for local efficiency loss.– Too many layers to be trended by human eyes.– Automated outlier detection script developed.
• Take out systematic effects that affect all layers in each tower.
• Take truncated (remove highest and lowest 10% of efficiencies) and weighted average of all runs for each layer.
• Tag instances of efficiencies if it is 5 away from the above average.
• Produce efficiency strip profile plots of outliers in comparison with the reference and save them as GIF for human inspection.
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 11
FMA (Bay 0) Layer X0, X2FMA (Bay 0) Layer X0, X2
• X0 and X2 do not have bad channels in these regions.– Shadow of bad strips
in X1 (1270-1500).– Inefficient regions are
slightly shifted.– Old efficiency calculation
is more susceptible.
Old efficiency calculation
X1 bad channel region
run ID (-77000000)
Rel
ati
ve
eff
icie
ncy
Eff
icie
ncy
strip ID
strip ID
Eff
icie
ncy
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 12
FM6 (Bay 12) Layer Y5FM6 (Bay 12) Layer Y5
• Noise flare according to Mutsumi.
run ID (-77000000)
Rel
ati
ve
eff
icie
ncy
Eff
icie
ncy
strip ID
Sugizaki
GLAST LAT Project LAT Instrument Analysis Meeting – May 26, 2006
Hiro Tajima, TKR Efficiency Trending 13
FM13 (Bay 7) Layer X15FM13 (Bay 7) Layer X15
• Another noise flare– Only 1st ladder is affected.– GTRC buffer limit is not
cause of inefficiency.– Probably due to dead time
of preamplifier.
run ID (-77000000)
Rel
ati
ve
eff
icie
ncy
Eff
icie
ncy
strip ID
Sugizaki