ANS RPSD 2014 - 18th Topical Meeting of the Radiation Protection & Shielding Division of ANS
Knoxville, TN, September 14 – 18, 2014, on CD-ROM, American Nuclear Society, LaGrange Park, IL (2014)
1
Digitalized two parametric system for gamma/neutron spectrometry
Martin Veškrna, Zdeněk Matěj, Filip Mravec, Václav Přenosil
Masaryk University, Botanická 15, Brno 612 00, Czech Republic
veskrna@mail.muni.cz
xmatejz@mail.muni.cz
xmravec@fi.muni.cz
prenosil@fi.muni.cz
František Cvachovec
University of Defence, Kounicova 65, 662 10 Brno, Czech Republic
frantisek.cvachovec@unob.cz
Michal Košťál
Research Center Rez ltd., 250 68 Husinec-Rez 130, Czech Republic
kostal@cvrez.cz
INTRODUCTION
Many types of detectors like stilbene, NE-213 etc. in
conjunction with photomultiplier loaded with low
working resistance produce pulses of approximately
100 ns length and contain information about deposited
particle in the trailing edge. Using fast analog to digital
converters (ADC) and field-programmable gate array
(FPGA) it is possible to create a spectrometric system
working in mixed gamma and neutron fields which is not
loaded dead time. The count rate of processed pulses can
reach more than one million per second.
Such a high count rate of processed pulses can be
achieved due to the pulse processing is implemented in
FPGA. The output of this pulse processing is amplitude
which describes the energy of deposited particle and
discrimination parameter whereby it is possible to
discriminate photons and neutrons. To increase the
dynamic range of energy of detectable particle the signal
from photomultiplier is separated into two branches with
different amplification. Each branch is digitalized by
separate ADC. Components from which the system is
composed are so light that the spectrometer can be easily
transported. Its weight is less than 3 kilograms.
Spectrometer was tested in the research reactor LR-0
in Rez near Prague (Czech Republic). The measured data
was processed using deconvolution into a neutron flux
density and compared with nowadays used analog
spectrometer and simulation result. Measured neutron
spectrum of Cf-252 is included.
THE AIMS OF DIGITAL SPECTROMETER
Neutron spectra in research reactor LR-0 in Rez are
nowadays measured by analog spectrometer which uses
photomultiplier with high working resistance load on
anode. [1] Therefore the length of measured pulses
approaching 1 millisecond. The information about
deposited particle is situated in leading edge of pulse. The
advantage of long pulses is low demands on evaluation
electronics because it is not necessary to work with high
frequencies. The disadvantage is a low count rate of
processed pulses which is up around 1000 pulses per
second.
The analog spectrometer processes separately the
amplitude of pulse and pulse shape discrimination
parameter. The digitalization occurs at the end of the
signal processing.
The problem of analog processing is a high
susceptibility to external noise and disturbances caused by
aging. The analog spectrometers are typically based on
Nuclear Instruments Modules (NIM) and therefore
contain many configurable components and connectors.
These components are due to the aging the source of
disturbance. The first aim of the digital spectrometer is to
minimize the number of components which are
susceptible to change the parameters due to aging,
especially analog adjustment elements.
The second aim of digital spectrometer is to increase
the count rate of processed pulses per second. Using the
low working anode resistance of photomultiplier the pulse
length is reduced approximately to 100 ns. The
information about the type of deposited particle is now
situated into the trailing edge of impulse – the
discrimination is still possible. [3]
The third aim is improving the resolution and pulse
shape discrimination at low energies. The signal from the
detector is separated into two branches which consist of
amplifier and analog to digital converter. Each branch has
different amplification.
2
Fig. 1. Scheme of two parametric digital spectrometer.
SPECTROMETER DESCRIPTION
The digital spectrometric system is built as a modular
system which allowing replacement of individual parts
according to the needs of the experiment (i.e. replacing
the amplifier when the detector is replaced). See Fig. 1.
The analog part consists only of the operating amplifiers
which adjust output signal of detector to input range of
ADC. All other signal processing is done in digital part.
Fig. 2. Two parametric digital spectrometer. From the
right: detector, amplifier with one input and two branches
with different amplification, digital spectrometer.
Amplifier
The analog amplifier splits the signal from detector
into two branches. Each branch is differently amplified
and digitalized by separate ADC. This different
amplification increases the dynamic range of energies of
particles that the spectrometer is able to process and
increase the signal to noise ratio. The ratio of
amplification of signal branches is about 8.
Both signal branches of amplifier do not change the
phase of the signal. Therefore it is possible to fuse the
signal at the beginning of digital processing and the future
processing is easier because it works only with one signal
path.
ADC card
The input analog signal is digitalized by two fast
ADC Texas Instruments ADC12D1000 working on
sampling frequency 1 GHz. Digital signal processing is
implemented in FPGA Xilinx Virtex-6. FPGA is able to
process all data flow from both ADC (24Gbits per
second). It means that the spectrometer is not loaded by
the dead time. Implemented discrimination algorithm
reduces the data flow to no more than several hundreds of
megabits per second.
It is possible to transfer raw data from ADC to
computer for future offline processing. This option is
ideal while designing and testing discrimination methods.
Computer
The computer is used to control the spectrometer and
for saving and processing of measured spectra. The
deconvolution of measured spectra is realized in the
computer. The connection between the ADC card and the
computer is realized by 10 Gbit optical ethernet. The high
speed is ideal while saving raw data from ADC.
PHOTON AND NEUTRON DISCRIMINATION
Digital spectrometer uses integration method which
is based on the principle of charge comparison (CC). The
principle of this method lies in comparison of area limited
by a trailing edge of the measured response with area
limited by the whole response. [3]
2
1
21 ,)(,)(
2
0
2
1
Q
Q
DdttiQdttiQ
t
t
t
t
 
Integrals Q1 and Q2 over the individual impulses can
be expressed numerically. The specific shape of impulses
depends on the used measuring apparatus. The result of
classification function D is a ratio between numerated
values of charges Q1 and Q2. Value t0 is always
determined by time of reaching maximum (amplitude)
and values t1 and t2 are set depending on the type of used
measuring apparatus. These are mainly a type and
parameters of the detector, photomultiplier parameters
and also range of energies of registered particles. Fig. 4
shows framework layout of these parameters.
For evaluation by suggested methods it is not
necessary to standardize impulse responses. The
suggested methods have linear computing complexity,
which is suitable to use for OFF-LINE as well as for ONLINE
evaluation. This method does not require any preprocessing
of the measured data.
Detector
Two branches amplifier
with different gain
FPGA Virtex-6
Computer
ADC 1
1 Gsample per second
ADC 2
1 Gsample per second
Signal
fusing
Pulse processing
- pulse shape discrimination
- amplitude subtraction
RAW data (24 Gbit per second)
1x
8x
10 Gbit
optical
ethernet
gamma/neutron
spectra spliting
deconvolution
optional RAW data
processing
3
Fig. 3. Neutron spectra in research reactor LR-0 and comparison with analog spectrometer and MCNP calculation.
Fig. 4. Framework layout of parameters of integration
method.
EVALUATION PROCESS
The presented digital spectrometric system was tested
in research reactor LR-0 in Rez near Prague (Czech
Republic). The neutron spectrum of Cf-252 is included.
Fig. 5. Neutron spectrum from spontaneous fission of
Cf-252.
The neutron spectra were measured in the energy
range from 1 to 10 MeV by the proton-recoil method
using a stilbene scintillator. The neutron spectra were
measured with a cylindrical stilbene detector of the
dimensions of 10x10 mm and 45x45 mm. The
deconvolution of the measured recoiled proton spectra
was done using the Maximum Likelihood Estimation. [2]
Fig. 5. shows the measured Cf-252 spectrum and
Table I. shows the spectral indexes. These are the ratios
between flux density in two energetic intervals. Based on
it is possible to compare the shape of two spectra. The
reference spectrum is described in [4].
Table I. Comparison of certain spectral indexes
Spectral index
(ratio between neutron flux
in two energetic intervals)
Digital
spectro-
meter
Spectral
indexes by
Sajo et. al.
[4]
(1 MeV, inf.) / (2 MeV, inf.) 1.70 1.67
(1 MeV, inf.) / (3 MeV, inf.) 2.99 2.97
(1 MeV, inf.) / (5 MeV, inf.) 10.21 10.18
(1 MeV, inf.) / (7 MeV, inf.) 36.44 38.22
(2 MeV, inf.) / (3 MeV, inf.) 1.76 1.77
(2 MeV, inf.) / (5 MeV, inf.) 6.01 6.08
(2 MeV, inf.) / (7 MeV, inf.) 21.45 22.85
Evaluation on LR-0 research reactor
Fig. 6. shows the LR-0 active zone arrangement. The
active zone consists of nine fuel cartridges and the
detector is situated in the center of the zone where there is
a high neutron flux. It causes a high count rate of detected
impulses and therefore it is possible to evaluate the
spectrometer on different count rate of processed
impulses.
Fig. 7. shows the discrimination data from the
spectrometer before the deconvolution process. The
neutron gamma discrimination is lower than 0.5 MeV.
0 1 2 3 4 5 6 7 8 9 10 11 12
Energy [MeV]
Fluxdensity[a.u.] MCNP calculation
Digital spectrometer - stilbene 45 x 45 mm - 420 000 pulses per second
Digital spectrometer - stilbene 10 x 10 mm - 40 000 pulses per second
Analog spectrometer - stilbene 10 x 10 mm - less than 1 000 pulses per second
4
Fig. 5. shows the measured neutron spectra and
comparison with measurement using analog spectrometer
which was done at the same place. [1] The result of
Monte-Carlo MCNP calculation is enclosed.
Fig. 6. Research reactor LR-0 active zone arrangement.
The zone is symmetrical and consists of 9 fuel cartridges.
The detector is situated in the center of the zone.
Fig. 7. Pulse shape discrimination data from 10x10 mm
stilbene (measured in LR-0 reactor). Photons are on the
left and neutrons on the right.
RESULTS
The presented results shows two parametric digital
spectrometric system designed to measure at high count
rate of pulses without dead time. This count rate can
exceed one million pulses per second. The spectrometric
system consists of fast analog to digital converters with
sampling frequency 1 GHz and FPGA. The pulse
discrimination is realized by integration method which is
based on principles of charge comparison methods.
The spectrometer was evaluated in research reactor
LR-0 in Rez near Prague. Measured count rate of pulses
exceeds 400 000 pulses per second. Additional evaluation
was carried out using Cf-252 neutron spectrum.
During the future improvement of digital
spectrometer it is necessary to focus on pulse
superposition. Due to high count rate of measured pulses
the probability of detection two or more pulses with small
time spacing increases. Then it is not possible to
discriminate it. Solving this problem allows the
measurement at count rate exceeding one million pulses
per second.
ACKNOWLEDGMENTS
The authors would like to acknowledge the support
of Technology Agency of the Czech Republic (project
TA01011383).
REFERENCES
1. Z. Bures, J. Cvachovec, F. Cvachovec, P. Celeda, B.
Osmera, “Multiparameter Multichannel Analyser System
for Characterization of Mixed Neutron – Gamma Field in
the Experimental Reactor LR-0”, Proc. Int. Conf. 11th
Symposium on Reactor Dosimetry, Brussels, Belgium, pp
194 – 201 (2002)
2. J. Cvachovec, F. Cvachovec, “Maximum Likelihood
Estimation of a Neutron Spectrum and Associated
Uncertainties”, Advances in Military Technology, Vol.1,
No. 2 January 2007 , pp. 5 – 28
3. G. F. Knoll, Radiation Detection and Measurement.
New York: Wiley, 2000, pp. 230–231.
4. E. Sajo, M. L. Williams, M. Asgari : Comparison of
measured and calculated neutron transmission through
steel for a 252 Cf source Ann. Nicl. Energy, Vol. 20, No.
9, pp. 285-304 (1993).
0 10 20 30 40 50 60 70 80 90 100
0
1
2
3
4
5
6
7
8
9
10
Pulse Shape Discrimination Parameter
NeutronEnergy[MeVee]