SiPM linearization status update Arjan Heering, Anton Karneyeu, Iouri Musienko, Mitchell Wayne HCAL Upgrade / 19 January 2017
SiPM SiPM is an array of small cells connected in parallel Cell release constant charge on hit Cells operate (“fire”) independently A cell needs time to recover the charge after “fire” (cell is ~not sensitive to new incoming photons until the end of recovery) – the source of non-linearity The output signal is a sum of signals produced by individual cells SiPM linearity is determined by its total number of cells
Setup Light source and path: - commercial LED pulser - Y11 scintillation fiber - clear fiber Photodetectors: - Production HCAL RM module (marked “RM1 not for P5”) with SiPM/QIE11 readout (QIE11 is not calibrated in this module) – non linear - Old RM module with HPD/QIE8 for light reference – linear Measured correction factor is: K = (HPD_charge / SiPM_pixels) vs. SiPM_pixels The setup is an update of light mixer measurements setup: - shorter LED pulse - HPD instead of APD for linear reference https://indico.cern.ch/event/525400/contributions/2151873/
Correction form Proposed correction form: K = p0 + p1*x + p2*x*x A = K * x x – SiPM pixels fired K – correction factor A – linearized output Previous talks: https://indico.cern.ch/event/562424/contributions/2273088/ https://indico.cern.ch/event/570231/contributions/2306446/
Dataset #1: 1 and 4 fiber, 2.8 and 3.3 mm: RAW - Ref. counts is HPD charge in fC - Sum of 10 time samples - Pedestal is subtracted
CORRECTION Spread < 4% up to ~300 GeV Plot is relevant up to 11k pixels 4-fiber channels get more light and go to higher number of pixels 10k pixels is approximately 300 GeV (in one depth segment) 3.3mm devices have 38k pixels total 2.8mm devices have 27k pixels total Spread < 4% up to ~300 GeV
CORRECTION: zoom to low light region Outliers: effect of QIE11 (un)calibration?
CORRECTION, no outliers Plot is relevant up to 11k pixels Spread < 2% up to ~300 GeV Statistics (number of curves) is low -> additional 1-fiber dataset with higher statistics
Dataset #2: 1 fiber, 2.8 and 3.3 mm: RAW
CORRECTION Spread < 3% up to ~300 GeV Relevant up to 14k pixels Outliers is same channels as dataset #1
CORRECTION, no outliers Relevant up to 14k pixels Spread < 1.5% up to ~300 GeV
CORRECTION, no outliers, global fit Correction curve is almost linear
Global fit: 2.8 mm devices Spread < 1.4% up to ~300 GeV Relevant up to 14k pixels NO. NAME VALUE ERROR 1 p0 1.00000e+00 fixed 2 p1 2.71238e-05 2.82432e-08 3 p2 1.32877e-10 2.01824e-12 Spread < 1.4% up to ~300 GeV
Global fit: 3.3 mm devices Spread < 1.4% up to ~300 GeV Relevant up to 14k pixels NO. NAME VALUE ERROR 1 p0 1.00000e+00 fixed 2 p1 2.59096e-05 4.41449e-08 3 p2 4.60721e-11 2.50893e-12 Spread < 1.4% up to ~300 GeV
Correction curves Old correction is based on light mixer measurements data https://indico.cern.ch/event/562424/contributions/2273088/ Correction factor SiPM pixels
Conclusion and plans Presented corrections are SiPM+QIE11, not only SiPM Correction form is almost linear The spread of correction curves < 2% up to ~300 GeV Plans: Measurements with calibrated RM/QIE11 Testbeam measurements
Backup
LED pulse shape
QIE11 calibration issue? A drop by ~4% on switch to subrange 2 point 3 / shunt=11.5 ~88pC/10.9 = 8.1pC range 3, subrange 1 point 0 / shunt=1 ~52pC range 4, subrange 1 point 2 / shunt=1 ~92pC range 4, subrange 2 point 1 / shunt=1 ~76pC range 4, subrange 2 A drop by ~4% on switch to subrange 2