Improvements to Readout
Electronics for the Compact Muon
Selenoid Hadron Calorimeter
Robert Schurz, IMSA;
Jacob Anderson, Fermilab
 Measures the energy of a
particle shower caused by
the collision of hadrons,
which are particles made
of quarks and gluons
such as pions, kaons,
protons, and neutrons
 Provides indirect
measurement of the
presence of uncharged
particles such as
neutrinos
2
CMS Hadron Calorimeter at CERN
Is a sampling calorimeter: finds the position,
energy, and arrival time of a particle
Uses alternating layers of brass absorbers and
fluorescent scintillators to calculate a
particle’s total energy
It attempts to capture every particle in the
particle shower caused by a proton collision
3
 When a particle hits a brass
absorber plate, an interaction
produces secondary particles
 Particles pass through
scintillators, producing a light
signal
 Optical fibers carry light signal
to hybrid photodiodes (HPDs)
4
 HPDs collect the light signals and convert them
into electronic signals by the photoelectric
effect that are amplified through high voltage
 The electrical signal is then digitized by a
charge integrating and encoding chip
 CMS wants to replace the old HPDs with
silicon multipliers
5
 Have low operating voltage, high gain, large
dynamic range, insensitivity to magnetic fields, and
good radiation tolerance (Mans et al., 2012)
 Large gain provides robust electrical signals that
reduce the importance of electrical noise (Buzhan et
al., 2002)
 The noise of the SiPM is negligible due to a high gain
of 6*104
compared to HPD gain of 2000 (Anderson,
2012)
 Increased effective quantum efficiency and ability to
detect single photoelectrons improves the ability to
calibrate and monitor the calorimeter (Crushman et
al., 1997) 6
 Absorption of a photon can excite an
avalanche that causes the struck pixel to
discharge
 Firing of a pixel causes its capacitor to
discharge, resulting in a quantized
charge output from the SiPM depending
on the number of pixels discharged
7
 Want to measure the charge distribution of each SiPM
 Used a simulation to measure the SiPM gain by two
methods:
 Light impulse from an LED
A photon causes a pixel to discharge
Measure gain by using sigma and the mean of a
Gaussian distribution for the charge distribution
assuming Poisson statistics for the number of
photons
 Electronic pedestal (PED) data
Thermal noise causes a pixel to discharge
Measure gain by subtracting the mean of the 0th
photoelectric peak from the mean of the 1st
pedestal
peak by using Gaussian distributions
How can we best measure
the gain of a SiPM: LED
amplitude or electronic
PED?
8
 Tested 158 mounting boards each with 18
different SiPM positions for each event lasting
50 ns
 Compared PED to LED method by looking at
PED gains, LED gains, difference in LED and
PED gains, and the LED signal amplitude
9
 Plot the charge distribution for the LED
amplitude and PED for a specific individual
SiPMs
 PED method shows a photoelectron peak and a
pedestal peak
 LED method had an amplitude that allowed us
to calculate the gain for each method
10
Figure 1. Charge Distribution for LED Amplitude (left) and PED
(right) for Mounting Board 58, Run Number 273, and SiPM Position
14.
11
 Found the root mean square of gains to be
0.3539 fC for the PED distribution and 0.9334
fC for the LED distribution
 Indicates that the PED method is a better
measurement technique for determining the
gain
12
Figure 2. Event vs. Gain for LED Amplitude (left) and PED (right).
Using the data from the charge distribution for each SiPM we were
able to calculate the gain. We obtained a RMS of 0.9334 fC for the
LED and 0.3539 fC for PED. 13
Figure 3. Events vs. Number of Photons for LED (left) and
Difference in LED and PED Gain (right). Using the Poisson
distribution we were able to calculate a mean of 586 photons for the
LED. Since the mean for the difference in LED and PED gain was
not centered at zero (systematic bias), we concluded that the LED
method had some error due to two photons hitting the same pixel.14
 PED method is a more robust measurement
technique because it causes less error
 The LED method had some saturation
 Quantified the electrical signal we get every time a
pixel fires which can then be converted back into the
energy that the particle deposited
 Future studies could measure:
 Noise
 Hamamatsu value + board offset
 Break-down voltage
 Dark current of the SiPM’s
 Other readout electronics such as QIE chip
15
The study was made possible by the collaboration of
IMSA and Fermilab efforts.
The author would like to thank all SIR staff
members, including Dr. Scheppler , and Jake
Anderson at Fermilab who made this investigation
possible.
16
Anderson, J. (2012). Upgrade of the CMS Hadron Calorimeter for an Upgraded LHC.
The Compact Muon Solenoid Experiment Conference Report, 1 (1), 234-240.
Buzhan, P., Dolgoshein, B., Ilyin, A., Kantserov, V. , Kaplin, V., Karakash, A.,
...Kayumov, F. (2002). An Advanced Study of Silicon Photomultiplier. Advanced
Technology & Particle Physics, 23 (2), 717-728.
Crushman, P., Heering, A., Nelson, J., Timmermans, C., Dugad, S. R., Katta, S., &
Tonwar, S. (1997). Multi-pixel Hybrid Photodiode Tubes for the CMS Hadron
Calorimeter. Nuclear Instruments and Methods in Physics Research, 27 (1), 107-112.
Mans, J., & CMS Collaboration (2012). CMS Technical Design Report for the Phase 1
Upgrade of the Hadron Calorimeter. The CMS Collaboration, 26 (3), 84-103.
17

Schurz FINAL presentation 5-02-13

  • 1.
    Improvements to Readout Electronicsfor the Compact Muon Selenoid Hadron Calorimeter Robert Schurz, IMSA; Jacob Anderson, Fermilab
  • 2.
     Measures theenergy of a particle shower caused by the collision of hadrons, which are particles made of quarks and gluons such as pions, kaons, protons, and neutrons  Provides indirect measurement of the presence of uncharged particles such as neutrinos 2
  • 3.
    CMS Hadron Calorimeterat CERN Is a sampling calorimeter: finds the position, energy, and arrival time of a particle Uses alternating layers of brass absorbers and fluorescent scintillators to calculate a particle’s total energy It attempts to capture every particle in the particle shower caused by a proton collision 3
  • 4.
     When aparticle hits a brass absorber plate, an interaction produces secondary particles  Particles pass through scintillators, producing a light signal  Optical fibers carry light signal to hybrid photodiodes (HPDs) 4
  • 5.
     HPDs collectthe light signals and convert them into electronic signals by the photoelectric effect that are amplified through high voltage  The electrical signal is then digitized by a charge integrating and encoding chip  CMS wants to replace the old HPDs with silicon multipliers 5
  • 6.
     Have lowoperating voltage, high gain, large dynamic range, insensitivity to magnetic fields, and good radiation tolerance (Mans et al., 2012)  Large gain provides robust electrical signals that reduce the importance of electrical noise (Buzhan et al., 2002)  The noise of the SiPM is negligible due to a high gain of 6*104 compared to HPD gain of 2000 (Anderson, 2012)  Increased effective quantum efficiency and ability to detect single photoelectrons improves the ability to calibrate and monitor the calorimeter (Crushman et al., 1997) 6
  • 7.
     Absorption ofa photon can excite an avalanche that causes the struck pixel to discharge  Firing of a pixel causes its capacitor to discharge, resulting in a quantized charge output from the SiPM depending on the number of pixels discharged 7
  • 8.
     Want tomeasure the charge distribution of each SiPM  Used a simulation to measure the SiPM gain by two methods:  Light impulse from an LED A photon causes a pixel to discharge Measure gain by using sigma and the mean of a Gaussian distribution for the charge distribution assuming Poisson statistics for the number of photons  Electronic pedestal (PED) data Thermal noise causes a pixel to discharge Measure gain by subtracting the mean of the 0th photoelectric peak from the mean of the 1st pedestal peak by using Gaussian distributions How can we best measure the gain of a SiPM: LED amplitude or electronic PED? 8
  • 9.
     Tested 158mounting boards each with 18 different SiPM positions for each event lasting 50 ns  Compared PED to LED method by looking at PED gains, LED gains, difference in LED and PED gains, and the LED signal amplitude 9
  • 10.
     Plot thecharge distribution for the LED amplitude and PED for a specific individual SiPMs  PED method shows a photoelectron peak and a pedestal peak  LED method had an amplitude that allowed us to calculate the gain for each method 10
  • 11.
    Figure 1. ChargeDistribution for LED Amplitude (left) and PED (right) for Mounting Board 58, Run Number 273, and SiPM Position 14. 11
  • 12.
     Found theroot mean square of gains to be 0.3539 fC for the PED distribution and 0.9334 fC for the LED distribution  Indicates that the PED method is a better measurement technique for determining the gain 12
  • 13.
    Figure 2. Eventvs. Gain for LED Amplitude (left) and PED (right). Using the data from the charge distribution for each SiPM we were able to calculate the gain. We obtained a RMS of 0.9334 fC for the LED and 0.3539 fC for PED. 13
  • 14.
    Figure 3. Eventsvs. Number of Photons for LED (left) and Difference in LED and PED Gain (right). Using the Poisson distribution we were able to calculate a mean of 586 photons for the LED. Since the mean for the difference in LED and PED gain was not centered at zero (systematic bias), we concluded that the LED method had some error due to two photons hitting the same pixel.14
  • 15.
     PED methodis a more robust measurement technique because it causes less error  The LED method had some saturation  Quantified the electrical signal we get every time a pixel fires which can then be converted back into the energy that the particle deposited  Future studies could measure:  Noise  Hamamatsu value + board offset  Break-down voltage  Dark current of the SiPM’s  Other readout electronics such as QIE chip 15
  • 16.
    The study wasmade possible by the collaboration of IMSA and Fermilab efforts. The author would like to thank all SIR staff members, including Dr. Scheppler , and Jake Anderson at Fermilab who made this investigation possible. 16
  • 17.
    Anderson, J. (2012).Upgrade of the CMS Hadron Calorimeter for an Upgraded LHC. The Compact Muon Solenoid Experiment Conference Report, 1 (1), 234-240. Buzhan, P., Dolgoshein, B., Ilyin, A., Kantserov, V. , Kaplin, V., Karakash, A., ...Kayumov, F. (2002). An Advanced Study of Silicon Photomultiplier. Advanced Technology & Particle Physics, 23 (2), 717-728. Crushman, P., Heering, A., Nelson, J., Timmermans, C., Dugad, S. R., Katta, S., & Tonwar, S. (1997). Multi-pixel Hybrid Photodiode Tubes for the CMS Hadron Calorimeter. Nuclear Instruments and Methods in Physics Research, 27 (1), 107-112. Mans, J., & CMS Collaboration (2012). CMS Technical Design Report for the Phase 1 Upgrade of the Hadron Calorimeter. The CMS Collaboration, 26 (3), 84-103. 17

Editor's Notes

  • #3 Particle interactions can tells us more about the formation of new particles such as the Higgs boson
  • #6 The light is converted to an electron via the photoelectric effect, but the amplification is through the high voltage providing enough kinetic energy to the photoelectron to allow it to create additional electron/hole pairs in the silicon sensor at the back of the HPD.
  • #8 Have pixels that are separated from the bias voltage by a current limiting quench resistor.
  • #9 For the LED method, we used a light pulse from an LED to cause photons to discharge pixels on the SiPM. We measured the gain by using the sigma and the mean of a Gaussian distribution for the charge distribution assuming Poisson statistics for the number of photons N. We assume that the uncertainty on N is √N. Since σ = gain * √N and Q = gain * N, we can use the relationship σ2/Q to calculate the gain for the SiPMs. For the electronic pedestal (PED) method, we left the circuit in the dark and allowed thermal noise to discharge a pixel. We then measured the gain by subtracting the mean of the 0th photoelectric peak from the mean of the 1st pedestal peak by using two Gaussian distributions to describe the peaks.