SiPMs A revolution for high dynamic range applications T. Bretz, - - PowerPoint PPT Presentation

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 1

SiPMs – A revolution for high dynamic range applications

• T. Bretz,
• J. Kemp, L. Middendorf,
• C. Peters, J. Schumacher
• R. Engel, R. Smida, D. Veberic

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 2

Mass Product → high precision → low cost product

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 3

Pierre Auger – AugerPrime SSD

Can SiPMs be used to read out the scintillator?

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 4

The SiPM optical module

• J. Schumacher, J. Kemp

Temperature feedback How identical are the two SiPMs? Is the binomial response understood?

2x 57,600 G-APDs

2 Lightguides 2 SiPMs (sum) 3 gain levels Power via USB

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 5

High precision

1 mm Example: Hamamatsu 1mm² SiPM

Geiger-mode avalanche photo diode

Credits: Hamamatsu

50µm

~2µm

Transistor in 2015: ~20nm(!)

Semi conductor photo sensors

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 6

Single photon counting

Credits: Hamamatsu

1pe 2pe 3pe 4pe 3pe 3pe 5pe 6pe 7pe 8pe

High precision → every avalanche (cell) releases similar charge

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 7

Binomial counting process

Fraction of maximum signal

• No. of impinging photons · PDE / number of cells
• No. of cells
• No. detected photons = No. of cells

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 8

Resolution decreases

→ Systematics become dominant!

→ Dynamic range increases!

Fraction of maximum signal

→ How identical are two devices of the same type?

• No. of impinging photons · PDE / number of cells

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 9

Compare two devices of same type

SiPMs Dark box

Rotating two SiPMs around a symmetry axis → After 180°, both SiPMs see the same light yield

LED ~10ns pulse

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 10

Charge state for measurement

→ Measurement close to saturation of sensor

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 11

Determine systematic error

LED

Fit: f(φ) = C + A · sin(φ - φ0)

LED

C = 0.7% ± 0.4% (syst.) φ = φ0 C Offset (syst. diff. of sensors) A Amplitude (syst. diff. of light distribution) φ0 Phase (setup dependent)

C = 0

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 12

Determine systematic error

LED

Fit: f(φ) = C + A · sin(φ - φ0)

LED

Consistent with 0.6% difference in gain due to different break- down voltage

φ = φ0 C = 0.7% ± 0.4% (syst.)

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 13

Determine systematic error

LED

Fit: f(φ) = C + A · sin(φ - φ0)

LED

Consistent with 0.6% difference in gain due to different break- down voltage

φ = φ0 C = 0.7% ± 0.4% (syst.)

→Two SiPMs of the same type are identical

if operated at identical over-voltage

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 14

The setup (Two SSD)

Experimental setup at KIT Selected high linearity Wavelength shifting fibers

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 15

The raw signal

Depends on history Is it understood?

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 16

Measured trace

• Measured trace is a convolution
• f the single-pe pulse

and the hit distribution

Amplitude / mV·0.8ns Measured trace Single PE Hit distribution ☼ single-pe

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 17

Deconvolution

Amplitude, PE / 0.16ns Measured signal Hit distribution Deconvolution

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 18

Hit distribution

• Deconvolve measured signal

into hit distribution

PE / 0.16ns Hit distribution

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 19

Simulating the SiPM response

• Deconvolve measured signal

into hit distribution

• Feed the trace into

a SiPM simulation

PE / 0.16ns Simulation input Simulation output

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 20

Matching the simulation output

• Deconvolve measured signal

into hit distribution

• Feed the trace into

a SiPM simulation

• Scale simulation

input until single-pe trace and simulation

• utput match

PE / 0.16ns, Photons / 0.16ns Simulation input Simulation output

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 21

The raw signal

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 22

The corrected signal

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 23

The corrected signal

~ x3 ~5%

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 24

Comparing two SiPM modules

SiPM 1 / SiPM 2 vs. PMT SiPM 1 vs. SiPM 2

→ 2nd SiPM module shows identical response

Thomas Bretz (RWTH Aachen University), Busan, South Korea, July 2017 25

Conclusion

• Largest systematic uncertainty is the

single-pe pulse (shape and amplitude)

• Simulation still under investigation
• Further measurements are foreseen

→ Two devices of the same type are identical within the precision

• f the applied over-voltage!

→ Proper processing increases the dynamic range significantly!