Professional Documents
Culture Documents
A Novel Technique For The Stabilization of SiPM Gain Against Temperature Varations PDF
A Novel Technique For The Stabilization of SiPM Gain Against Temperature Varations PDF
I. INTRODUCTION
Manuscript received June 07, 2012; revised September 24, 2012; accepted
February 15, 2013. Date of current version April 10, 2013.
F. Licciulli and C. Marzocca are with the Department of Electrical and
Electronic Engineering, Politecnico di Bari, I70125 Bari, Italy (e-mail:
marzocca@poliba.it).
I. Indiveri is with the Italian National Institute of Nuclear Physics (INFN),
Sezione di Bari, Bari, Italy (e-mail: ivanoindiveri@libero.it).
Digital Object Identifier 10.1109/TNS.2013.2249527
As pointed out above, the main cause of the gain drift with
the temperature for a SiPM is the temperature dependence of the
breakdown voltage, which is expressed by the following equation [7]:
(2)
607
Fig. 1. Schematic architecture of a classic compensation scheme for the temperature dependence of SiPM gain.
the thermistor in one of its branches, is able to track the variations of the breakdown voltage, thus making the gain constant.
We propose an alternative solution, based on the direct measurement of the temperature dependence of the SiPM gain. Basically, let us consider a SiPM of the same kind of those used
in the considered application as light sensors and shield it from
any exposure to external photons. This blind SiPM will generate only dark pulses with amplitude approximately proportional to its gain. In fact, the resulting waveform associated to a
single dark pulse can be approximated as the response of a linear
system to a very short current pulse, i.e., a Diracs delta
,
which describes the current generated by the triggering of an
avalanche breakdown [12]. If the blind detector is in thermal
equilibrium with the sensitive devices, the amplitude of its dark
pulses will track the temperature variations according to the
temperature dependence of the SiPM gain [9].
In the proposed compensation scheme, the average value of
the dark pulse amplitude
is measured over an appropriate time interval in which the temperature can be considered
constant. Thus, a negative feedback loop is established around
the blind detector: the difference between
and a desired
reference value
is amplified and superimposed to the bias
voltage of the detector, as depicted schematically in Fig. 2. If the
loop gain is adequately large, in presence of temperature variations, the bias voltage is automatically adjusted by the feedback
loop so that the amplitude of the dark pulses is kept constantly
equal to the reference value, hence also the gain of the blind
SiPM does not vary. The same bias voltage variations generated by the feedback loop for the blind device are also applied
to the sensitive SiPMs, thus achieving the desired gain control.
For example, in the ideal arrangement shown in Fig. 2 the
amplitude of the dark pulses generated by the blind SiPM is
extracted by means of a peak detector and averaged by a lowpass filter, to obtain
.
If this kind of approach is used, the advantages of negative
feedback are fully exploited and there is no need of accurate
measurements of the temperature sensitivity of the breakdown
voltage, which is variable considering different SiPM manufacturers and even different lots of SiPMs of the same type. Moreover no special control of the amplifier gain is necessary and
the influence of parametric drifts of the system components can
be made negligible, thus avoiding the need of long and complicated calibration procedures, which involve an accurate control
of the temperature. The only requirement is that the blind SiPM
used as a temperature (or gain) sensor must be matched to the
detectors used as light sensors in the specific application.
Since the time constants of the temperature variations to be
compensated are usually very slow, no particular requirements
608
Fig. 3. Block diagram of the experimental setup used to prove the effectiveness
of the proposed temperature compensation principle.
which is proportional to
, i.e., to the SiPM gain. Thus, if
is conveniently increased, a good indicator of the SiPM gain is
still obtained also in presence of optical cross talk. Similar considerations can be done for what concerns afterpulsing, which,
by the way, affects the charge associated to a dark pulse much
more than its peak amplitude.
Since, in each of the measurement cycles of our setup, the bias
voltage can be varied only by steps
, to appreciate the
corresponding variation of the dark pulse amplitude
609
(where is a constant), in presence of all the mentioned sources of noise, the parameter should be set so that a
given signal to noise ratio SNR is guaranteed:
where
is the total variance of the measurement error
which affects
. As a consequence the minimum number
of samples to be considered should be:
(6)
Thus, if
is reduced to decrease
, the number
of samples
should be increased according to (6). In practice, once
has been set, the parameter
must be increased until each measurement cycle provides the same final
value
, at constant temperature.
Two 1 mm 1 mm SiPM of the same type, manufactured
by FBK-irst and composed by 400 micro-cells 50 m 50 m,
have been used in the experiments, one as blind SiPM and the
other as sensitive detector. The gain of the sensitive SiPM as
a function of the temperature has been also monitored via the
amplitude of its own dark pulses. Both detectors have been enclosed in a thermally isolated dark box and their temperature has
been controlled by means of a Peltier cell in the interval between
20
and 30 . Two AC-coupled transimpedance preamplifiers, with gain of about 2
and bandwidth of 20 MHz, based
on discrete BJTs, have been used to read-out the detectors. Temperature control via the Peltier cell, instrument control via GPIB
interface and all the data transfer and processing operations have
been managed by software routines which run on the same host
PC.
Since, due to limited resolution in the control of the Peltier
cell we have used, the minimum temperature variation which
can be reliably imposed to the system is
, (5)
yields
, knowing that
.
First, the temperature has been kept constant at
and a long series of measurement cycles have been carried out,
i.e., a series of ramps of
have been applied to the blind
SiPM, to find a suitable value for the parameter . According to
the previous considerations, the value of has been increased
until a constant
has been achieved, yielding
. In
Fig. 4 the values of
obtained considering the first ten
measurement cycles are shown. In these conditions, an average
value of 5.4 mV has been measured for
, with a standard deviation of 145
, so that the relative variation is about
2.7%. The amplitude of the dark pulses generated by the sensitive SiPM has been also monitored and similar results have been
obtained.
Considering Fig. 4, note that a complete measurement cycle,
which requires the acquisition of about 500 dark pulses, takes
about
to be completed, due to the speed limitations
introduced by the GPIB interface used to transfer the acquired
data from the oscilloscope to the PC. As a consequence, temperature variations slower than
can
be efficiently tracked by the system.
Fig. 4. Average amplitude of the dark pulses generated by the blind SiPM ob.
tained considering 10 measurement cycles at constant temperature
Fig. 5. Average amplitude of the dark pulses generated by the blind SiPM as a
function of its bias voltage, for each considered value of the temperature.
610
Fig. 7. Average amplitude of the dark pulses generated by the sensitive SiPM
as a function of the temperature, with (closed loop) and without (open loop)
temperature compensation.
used as temperature sensors within the proposed compensation scheme. Furthermore, alternative implementations of the
technique can be based on direct measurements of the detector
gain, i.e., of the average value of the charge contained in the
dark pulses generated by the blind SiPM: this approach could
results in complication of the required circuitry, but it could
possibly guarantee more accuracy.
ACKNOWLEDGMENT
The authors would like to thank Dr. C. Piemonte, from
FBK-irst, for providing the SiPM samples used in the experiments, and prof. F. Corsi, DEE Politecnico di Bari, for helpful
discussions.
REFERENCES
[1] G. Bondarenko et al., Limited Geiger-mode microcell silicon photodiode: New results, Nucl. Instr. Meth. Phys. Res. Sec. A, vol. A 442,
pp. 187192, 2000.
[2] P. Buzhan et al., Silicon photomultiplier and its possible applications, Nucl. Instr. Meth. Phys. Res. Sec. A, vol. A 504, pp. 4852,
2003.
[3] C. Piemonte, A new silicon photomultiplier structure for blue light detection, Nucl. Instr. Meth. Phys. Res. Sec. A, vol. A 568, pp. 224232,
2006.
[4] V. Golovin and V. Saveliev, Novel type of avalanche photodetector
with Geiger mode operation, Nucl. Instr. Meth. Phys. Res. Sec. A, vol.
A518, pp. 560564, 2004.
611
[5] J. C. Jackson, P. J. Hughes, D. Herbert, A. Stewart, and L. Wall, Review of SPM low light level detectors, in Proc. 19th IEEE Lasers
Electro-Optics Soc. Meeting, 2006, pp. 723724.
[6] C. Y. Chang, S. S. Chiu, and L. P. Hsu, Temperature dependence of
breakdown voltage in silicon abrupt P-N junctions, IEEE Trans. Electron Devices, vol. 18, no. 6, pp. 391393, 1971.
[7] M. Petasecca et al., Thermal and electrical characterization of silicon
photomultiplier, IEEE Trans. Nucl. Sci., vol. 55, no. 3, pp. 16861690,
2008.
[8] M. Ramilli, Characterization of SiPM: Temperature dependencies,
in Proc. 2008 IEEE Nucl. Sci. Symp. Med. Imag. Conf. Rec., 2008, pp.
24672470.
[9] R. Bencardino and J. E. Eberhardt, Development of a fast-neutron
detector with silicon photomultiplier readout, IEEE Trans. Nucl. Sci,
vol. 56, no. 3, pp. 11291134, 2009.
[10] P. S. Marrocchesi et al., Active control of the gain of a 3 mm 3
mm silicon photoMultiplier, Nucl. Instr. Meth. Phys. Res. Sec. A, vol.
A602, pp. 391395, 2009.
[11] H. Miyamoto, M. Teshima, B. Dolgoshein, R. Mirzoyan, J. Nincovic,
and H. Krawczynski, SiPM development and application for astroparticle physics experiments, in Proc. 31st Int. Cosm. Ray Conf., Lodz,
Poland, 2009.
[12] F. Corsi et al., Preliminary results from a current mode CMOS
front-end circuit for silicon photomultiplier detectors, in Proc. IEEE
Nucl. Sci. Symp. Med. Imag. Conf. Rec., 2007, pp. 360365.
[13] S. Callier, F. Dulucq, R. Fabbri, C. de La Taille, B. Lutz, G. MartinChassard, L. Raux, and W. Shen, Silicon Photomultiplier Integrated
Readout Chip (SPIROC) for the ILC: Measurements and possible further development, in Proc. 2009 IEEE Nucl. Sci. Symp. Med. Imag.
Conf. Rec., 2009, pp. 4246.
[14] F. Corsi, M. Foresta, C. Marzocca, G. Matarrese, and A. Del Guerra,
CMOS analog front-end channel for silicon photo-multipliers, Nucl.
Instr. Meth. Phys. Res. Sec. A, vol. A617, pp. 319320, 2010.