Journal of Analytical Sciences, Methods and Instrumentation, 2013, 3, 152-157
http://dx.doi.org/10.4236/jasmi.2013.33018 Published Online September 2013 (http://www.scirp.org/journal/jasmi)
The Basis for Design and Manufacture of a DSP-Based
Coincidence Spectrometer
Nguyen Xuan Hai1*, Pham Ngoc Tuan1, Nguyen Nhi Dien1, Dang Lanh1, Tuong Thi Thu Huong1,
Pham Dinh Khang2
1Nuclear Research Institute, Dalat, Vietnam; 2Nuclear Training Center, Hanoi, Vietnam.
Received July 1st, 2013; revised August 8th, 2013; accepted August 17th, 2013
Copyright © 2013 Nguyen Xuan Hai et al. This is an open access article distributed under the Creative Commons Attribution Li-
cense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ABSTRACT
The coincidence technique and the coincidence spectroscopy have been developed and applied for over 40 years. Most
of popular coincidence measurement systems were based on analog electronics techniques such as time to amplitude
conversion (TAC) or logic selecting coincidence unit. The above-mentioned systems are relatively cumbersome and
complicated to use. With the strong growth of digital electronics techniques and computational science, the coincidence
measurement systems will be constructed simpler but more efficient with the sake of application. This article presents
the design principle and signal processing of a simple two-channel coincidence system by a technique of digital signal
processing (DSP) using Field Programmable Gate Arrays (FPGA) devices at Nuclear Research Institute (NRI), Dalat.
Keywords: Coincidence Spectrometer; Digital Signal Processing; FPGA
1. Introduction
The background reduction measurement systems with
active methods are based on coincidence or anti-coinci-
dence techniques. They are mainly built from proper
functional electronics modules in NIM or CAMAC stan-
dards [1,2]. They allow us to identify coincidence or anti-
coincidence events via main electronics blocks called
coincidence unit or time to amplitude converter. Nor-
mally, basic configuration of a coincidence measurement
system will, at least, consist of two channels, and the
selection of coincidence event pairs depends on the de-
fining moment of the pulses appearing at “Timing” out-
put of the pre-amplifier. Obvious drawback of this sys-
tem is cumbersome in size, adjusting operation and syn-
chronizing signals among electronics stages.
The growth of computer engineering and programma-
ble devices that are capable of operating at high frequen-
cies has allowed us to design a new spectrometry genera-
tion. The spectrometry generation is compact on size,
simple in terms of connectivity and using [2-4].
At Nuclear Research Institute (NRI), Dalat, a number
of researches and construction of Compton suppression
as well as event-event coincidence systems were perfor-
med in 1990s. The results of these studies were reported
in [5]. In the recent period at NRI, further studies on the
coincidence spectrometer were presented in several pub-
lications [6-8]. Overall, although there have been sig-
nificant improvements on acquiring as well as processing
data, this system is still based on the traditional way in
obtaining signals under the operation of a coincidence
unit or TAC. With the research results gathered during
the installation, investigation and exploitation of an
“event-event” coincidence spectrometer on digital signal
processing (DSP) at NRI, we have now proposed a new
approach for a design of the DSP-based multi-application
coincidence technique through Field Programmable Gate
Arrays (FPGA) devices. The basis for the design of this
spectrometer will be presented within the framework of
this article.
2. The Basis and Method for the Design
2.1. Design of Multi-Channel Analyzer
The block diagram of MCA is presented in Figure 1. The
functions of the diagram can be summarized as follows:
The pre-filter works as high-pass filter giving an out-
put signal with a constant shaping time. In addition, the
pre-filter has amplification function to generate the ap-
propriate signals for ADC conversion.
The pre-filter’ output signals are sent to flash ADC for
*Corresponding author.

The Basis for Design and Manufacture of a DSP-Based Coincidence Spectrometer 153
Figure 1. The MCA block diagram.

The Basis for Design and Manufacture of a DSP-Based Coincidence Spectrometer 154
sampling. The ADC’ output signals are digital copies of
input’ analog signals.
The transfering “Exponential Decay to Rectangular
Shape” circuit converts exponential decay to rectangular
pulse, and its function is the opposite of high pass de-
convoler (HPD). The transfering “Rectangular Shape to
Trapezoidal Shape” circuit converts shape of rectangular
to trapezoidal pulse, and its function is the low pass filter
(LPF). The circuits of pulse pile-up rejection, base line
restoration and built-in configuration are also designed in
the main board.
The pulse pile-up rejection circuit detects the pile-up
pulses in duration from rise time to half of the flat width,
in slow channel. In case of non-overlapped pulse is de-
tected in monitoring duration, the right pulse will be
analysis. A dual port memory is integrated in FPGA de-
vice for buffering data. The preset time is set by user
with capacity up to 4.2 × 109 seconds. The interface cir-
cuit is integrated in the main board and connected to PC
through a USB-RS232 bridge.
2.2. Design of the “Event-Event” Coincidence
Spectrometer
The block diagram of the “event-event” coincidence
spectrometer is presented in Figure 2. According to the
block diagram, the coincidence spectrometer consists of
three parts; they are two DSP-based multi-channel ana-
lyzers and one channel of timing discriminators. Two
MCA channels are used for getting the energy spectrum
from the two detectors. Timing discrimination circuits
are used for the time interval measurement. The pre-filter,
Fast ADC and DSP MCA are the components of the en-
ergy analysis channel. The FAST COMPARATOR,
TIME COUNTER, TIME COMPARATOR and COIN-
CIDENT CONTROLLER are the components of the
coincidence channel.
Except for the PRE-FILTER, FAST ADC and DDR2
MEMORY, almost other components are designed and
integrated into one FPGA device which is named
XC6SLX16.
Figure 2. The block diagram of the “event-event” coinci-
dence spectrometer.
The external DDR2 MEMORY is used to save the
measured results.
The interface circuit is integrated on the main board
and connected to PC through USB-RS232 bridge.
Principal Operation
Two MCAs receive pulses from the preamplifier’s en-
ergy outputs, the pulses’ amplitudes are analyzed (by
DSP techniques); simultaneously, the preamplifier’s
output pulses are connected to fast comparator circuits.
The output signals of these comparator circuits are used
to start/stop a time counter. The used clock frequency is
400 MHz (2.5 ns period). The contents of the counter
(time interval between two events) are compared with
values of preset time interval; if the measured time is in
the selected range, the amplitudes of pair of events will
be recorded. Three values E1, E2 and T will be stored
into memory; in which E1 and E2 are energy values of
pair of coincidence gamma rays, and T is time interval.
The measured results are read by data acquisition pro-
gram and saved to file on the hard disk.
3. Fabricating
3.1. Hardware
The entire design is programmed and configured in the
FPGA chip XC6LX16-CS324. The used main board is
SP601 kit supplied by Xilinx as Figure 3.
3.2. Data Acquisition Program
The data acquisition program for energy and timing
spectra was written under LabWIEW 8.5. The functions
of program are as follows:
+ Connecting peripheral devices to PC and interfacing
to PC through USB port.
+ Control of data acquisition for multi-MCA mode:
start/stop data acquisition, preset measurement time, data
saving…
+ Control of data acquisition for “event-event” coin-
cidence mode: start/stop data acquisition, preset meas-
urement time, data saving…
+ The base data analysis: display spectra, counts per
channel, energy calibration, zoom in and zoom out spec-
tra.
3.3. The Main Features of Spectroscopy
1) The multi-channel analysis:
- Input: accept the preamplifier’s output pulses from
semiconductor detector, the amplitude from tens to 500
mV;
- Resolution: 8192 channels;
- Capacity of counts/channel: 232 − 1;
- Preset time capacity: 232 − 1 seconds with 1 second

The Basis for Design and Manufacture of a DSP-Based Coincidence Spectrometer 155
Figure 3. The main board.
step;
- The integral nonlinearity: <±0.05%;
- Shaping time: 0.4; 0.8; 1.6; 3.2; 6.4; 12.6; 25.2 μs;
- Coarse gain: 1, 2, 5 and 10;
- Fine gain: 0.75 ÷ 1.25, step 0.01;
- Displaying and rejecting the pile-up pulses, baseline
correction and dead time correction.
2) Event-event coincidence:
- Preset coincidence time window: 5 ÷ 10,000 ns, res-
olution step 5 ns;
- Mode: coincidence, PHA;
- Memory capacity: 16 M events.
3) Power supply:
+12 V/800 mA,
−12 V/50 mA,
+6 V/300 mA,
−6 V/100 mA.
4) Dimensions: width NIM 2M.
4. Testing and Results
The experiment configuration is showed in Figure 4. The
disk-shaped 60Co source with activity of 24 kBq was
placed between the two detectors (GMX35). The dis-
tance from source to the detectors’ surfaces was 4 cm. In
front of each detector, a sheet of lead 2 mm in thickness
Figure 4. Experiment configuration.
was placed to reduce the effect of back scattering from
the detectors to each other.
Data were collected during 2 h. The interfacing pro-
gram was set up in event-event coincidence mode. The
data files were saved to hard disk automatically in all
4096 coincidence events. The Figures 5-7 are energy
spectra of each channel, summation spectra and timing
spectra respectively.
Figure 6 shows the 60Co summation spectra, with only
one peak and Compton scattering background. The coin-
cident spectrum (Figure 5) of each channel contains two

The Basis for Design and Manufacture of a DSP-Based Coincidence Spectrometer 156
Figure 5. Energy spectrum of 60Co.
Figure 6. Summation spectra of 60Co.
Figure 7. Timing spectra (one clock unit equivalent 2.5 ns).
peaks (1173 and 1332 keV) as in MCA mode with gate
control. The measured energy resolution of 1332 keV
peak was ~8.5 keV, which was worse than three times in
comparison with detectors’ nomination value.
Figure 7 shows the timing spectra measured with 60Co
source. The time resolution and the peak/background
height ratio were ~12.5 ns and 2453:220 respectively.
The timing resolution was good but the spectrum shape
was not smooth.
Figure 6 shows only the full-energy peak of
 -rays of
cascade. In practice, the nuclide 60Co decay
 -emission
to 2.5 MeV level in 60Ni. The 2.5 MeV level is de-excited
by a cascade of 1.17 and 1.33 MeV
 -rays to the ground
state. None of cross-over is observed. This showed that
the data acquisition algorithm and program for DSP co-
incidence spectrometer are exact but the energy resolu-
tion and timing spectra are not good.
5. Conclusions
The design of systems based on DSP techniques using
FPGA allows constructing a simple, compact and impact
coincidence spectrometer in which all of parameters are
selected and controlled by software.
Sample frequency is very important and reduced noise
is a big problem. These are problems that should be
solved in designing and manufacturing DSP coincidence
spectrometer.
Currently, the reduced noise for DSP coincidence
spectrometer has been studying at the Department of
Nuclear Physics and Electronics, NRI, Dalat. It is hope-
ful that, in the near future, the design of system might be
the basis for development and application of coincidence
measurement techniques in the field of physics research
and applications.
REFERENCES
[1] V. T. Jordanov and G. F. Knoll, “Digital Synthesis of
Pulse Shape in Real Time for High Resolution Radiation
Spectroscopy,” Nucle ar Instruments and Methods in Phy-
sics Research Section A, Vol. 345, No. 2, 1994, pp. 337-
345.
[2] M. Lauer, “Digital Signal Processing for Segmented
HPGe Detectors Preprocessing Algorithms and Pulse
Shape Analysis,” Doctor of Science Thesis, University of
Heidelberg, Heidelberg, 2004.
[3] W. R. Leo, “Techniques for Nuclear and Particle Physics
Experiments,” Springer-Verlag, Berlin, Heidelberg, 1987.
[4] Valentin T. Jordanov and Glen F. Knoll, “Digital Tech-
niques for Real-Time Pulse Shaping in Radiation Meas-
urements,” Nuclear Instruments and Methods in Physics
Research Section A, Vol. 353, No. 1-3, 1994, pp. 261-
264.
[5] http://www.fastcomtec.com/products/multiparameter-mul
tichannel-analyzers/mpa-3rtc.html
[6] A. Kimura, Y. Toh, M. Koizumi, A. Osa, M. Oshima, J.
Goto and M. Igashira, “Development of a Data Acquisi-
tion System for a Multiple Gamma-Ray Detection Me-
thod,” AIP Conference Proceedings, Vol. 769, Melville,
New York, 2005.
[7] A. M. Hoogenboom, “A New Method in Gamma-Ray
Spectroscopy: A Two Crystal Scintillation Spectrometer
with Improved Resolution,” Nuclear Instruments, Vol. 3,

The Basis for Design and Manufacture of a DSP-Based Coincidence Spectrometer 157
No. 2, 1958, pp. 57-68.
doi:10.1016/0369-643X(58)90092-6
[8] A. Pullia, “Quasi-Optimum Gamma and X-Ray Spectros-
copy Based on Real-Time Digital Techniques,” Nuclear
Instruments and Methods in Physics Research A, Vol.
439, No. 2-3, 2000, pp. 378-384.
doi:10.1016/S0168-9002(99)00897-9