Journal of Analytical Sciences, Methods and Instrumentation, 2013, 3, 158-162
http://dx.doi.org/10.4236/jasmi.2013.33019 Published Online September 2013 (http://www.scirp.org/journal/jasmi)
A Design Configuration of an FPGA-Based Coincident
Spectrometry System
Pham Dinh Khang1, Nguyen Nhi Dien2, Dang Lanh2, Nguyen Xuan Hai2, Pham Ngoc Tuan2,
Nguyen Duc Hoa3, Nguyen An Son3
1Nuclear Training Center, Hanoi, Vietnam; 2Nuclear Research Institute, Dalat, Vietnam; 3University of Dalat, Dalat, Vietnam.
Received July 8th, 2013; revised August 9th, 2013; accepted August 16th, 2013
Copyright © 2013 Pham Dinh Khang 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
In the past, most of popular coincidence spectrometers were normally based on traditional electronics techniques such
as time to amplitude conversion or logic selecting coincidence unit. They were complicated and it is not convenient for
us to use them. This paper deals with a new design of a contemporary coincidence spectrometer which is based on Field
Programmable Gate Arrays (FPGA) devices via Digital Signal Processing (DSP) techniques with Hardware Description
Language (VHDL). The outstanding advantage of DSP techniques and FPGA technology is capable of enhancement of
the quality of the experimental measurements for nuclear radiation. The designed configuration of the traditional system
was tested on the PCI 7811R board of National Instruments while the digital systems were establishing with FPGA de-
vices. The purpose of this work is referring to the principle for construction of an FPGA-based system capable of re-
placing a conventional system. Therefore, a novel approach for in-house development of digital techniques is presented.
The method for designing the system is utilization of slow-fast coincidence configurations with two HPGe detectors
obtaining a pair of coincidence events, processing data in DSP algorithms. The significant and noticeable results are the
operating frequency of 80 MHz and system timestamp window of approximately 10 ns.
Keywords: Digital Signal Processing; FPGA; VHDL
1. Introduction
A spectrometry system is used to measure nuclear radia-
tion including X-rays, gamma rays, alpha and beta rays,
neutrons and other heavy charged particles. Most of these
applications require the identification of radiation sour-
ces and relative concentration of each source. The ex-
perimental measurements in nuclear physics may include
counting events, getting information about energy-timing,
and their combinations. Typically, these main parameters
are changed in the measurement of radiation energy
range and speed of the whole event. Decay scheme and
lifetime studies, coincidence experiments, single-photon
counting, and position annihilation studies are some of
the experimental areas that require good timing capabili-
ties [1]. As well-known, all of the traditional spectrome-
try systems were not much convenient because of com-
bining properly functional electronics modules in the
Nuclear Instrumentation Modules (NIM) or the Com-
puter Automated Measurement and Control (CAMAC)
standards [2,3]. In the two-past decades, at Nuclear Re-
search Institute (NRI), Dalat, a number of researches and
construction of Compton suppression as well as event-
event coincidence systems were carried out with suitable
results reported in Ref. [4]. Recently, some of additional
investigations about the coincident spectrometry systems
were issued in several works [5]. Although these systems
had been operating well, they were still conventional
electronics models; therefore, they should be replaced by
contemporary electronics techniques. One of the new
development directions for building experimental sys-
tems of nuclear physics studies and applications of nu-
clear technology is utilization of FPGA and DSP tech-
niques. This direction meets effectively the more in-
creasing requirements on the accuracy of ionizing radia-
tion measurements. Since that, a novel generation of
spectrometry systems (coincidence or anticoincidence
mode) is compact on size, convenient in terms of con-
nectivity and use [6]. The outstanding advantage of DSP
techniques and FPGA technology is capable of en-
hancement of the quality of the experimental measure-
ments for nuclear radiation, minimization of functional

A Design Coinfiguration of an FPGA-Based Coincident Spectrometry System 159
electronics modules as well as the economic investment
[7,8]. Besides, an important element of the system based
on DSP and FPGA is low power consumption to save
energy that has a special meaning in large equipments.
With these advantages, the applied research via FPGA,
DSP in design and fabrication of radiation measurement
instruments for fundamental research in nuclear physics,
especially about the study of nuclear structure and data
on neutron beams at the Dalat reactor and on the charged
particle beam accelerators for domestic needs is essential.
2. The Basis and Method for the Design
2.1. The Basis for Operation of an
“Event-Event” Recording Coincidence
Spectrometer Using TAC
An “event-event” recording coincidence spectrometer for
processing data under a combination of the traditional
analog electronics and interfacing unit is shown in Fig-
ure 1. The standalone spectrometer is capable of linking
to PC through a Peripheral Component Interconnect (PCI)
slot. The system consists of two channels: the upper
channel and the lower one. All of the functional elec-
tronics modules are as follows:
 Spectroscopy amplifier (AMP), model 572A, Can-
berra;
 Fast peak sensing Analog to Digital Converter-(ADC)
8 k, model 7072, FAST CompTec;
 High Voltage (HV) bias supply, model 660, Ortec;
 Timing Filter Amplifier (TFA), model 474, Ortec;
 Constant Fraction Discriminator (CFD), model 584,
Ortec;
 Delay unit, model 2058, Canberra;
 Time to Amplitude Converter (TAC), model 566,
Ortec;
 Peak sensing ADC 16 k, model 8713, Canberra.
The principal operation of the system is as follows
(refer to Figure 1): The signals appearing on the energy
(E) outputs from two HPGe detectors are fed to both the
inputs of two AMPs belonging to the upper and lower
channels. In addition, two timing (T) outputs are also
coming to the inputs of two TFAs. Next, the two-TFA
outputs are fed to the inputs of CFDs. The upper CFD
output will then strobe the “Start” input TAC, and the
other CFD output plays a role of stopping the TAC in
one converting cycle for changing signal amplitude.
When the interfacing unit 7811R receives “Valid conver-
sion” signal from TAC, it will soon send back three
“Gate” signals for gating all of ADCs; and at that time,
ADCs are allowed to convert amplitude into BCD code
numbers. After ADCs finish conversion cycles, the in-
terfacing unit will read the code and write these values
into memory.
After finishing write operation, ADCs return to an idle
status and wait for another gate signal under the control
of the next valid conversion signal. ADCs have no opera-
tion if there is no strobe signal at the gate regardless of
incoming signals at their inputs. In a data file, the data
will be arranged into three columns E1(n), E2(n) and
E3(n). The values of E1(n) and E2(n) are in turn the am-
plitude code numbers of the two coincidence pulses
coming from detector 1 and detector 2 respectively, E3(n)
is a value corresponding to the differential time between
two events, n is the ordinal number of pairs of coinci-
dence events from the beginning of measurement. After
finishing the measurement, data is handled by the multi-
Figure 1. The block diagram of TAC-based coincidence spectrometer used at NRI [5].

A Design Coinfiguration of an FPGA-Based Coincident Spectrometry System 160
variable statistics processing program to obtain informa-
tion about energy, transition intensity and decay scheme
of nuclei on research. Dead time (DT) of the system will
be calculated as the shortest interval between two pairs of
consecutive amplitude codes recorded via spectrometer.
Total DT of the system depends on the sampling rate of
ADC and data-transfer speed of interfacing unit. The
slower they work, the longer dead time, and vice versa.
Because DT is one of the causes affecting the system’s
efficiency, therefore it should be ensured that the shorter
the DT, the higher the data.
min1 2 34

  
where τ1: delay of spectroscopy amplifier, τ2: shaping
time, τ3: ADC’s conversion time, and τ4: time for data
transfer operation of the interfacing unit.
2.2. The Principle for Design of an
“Event-Event” Recording Digital
Coincidence Spectrometer
Figure 2 shows the block diagram of an “event-event”
recording digital coincidence spectrometer. Its operation
principle is as follows: when the radiation signals are
recorded from detector 1 or detector 2, DSPs analyze the
amplitude of the pulses and then give the corresponding
values of (A1, A2). At the same time, when the signals
exceed the lower thresholds, DSPs will read more the
values corresponding to the moments of (t1, t2) at those
the pulses exceed the aforementioned thresholds. The
timing tester will determine the time difference between
the two events Δt = |t1 – t2|. If ΔW is called coincidence
time window of the system, there are a number of cases
occurred as follows:
Δt ≤ ΔW: coincidence occurred. The programmer will
write pair of events into memory with contents of A1, A2
and Δt.
Δt > ΔW: non-coincidence occurs. The program will
remove pair of events.
In the second case, assuming that there was an event
appeared beforehand on the first channel, the program
will remove the values of A1 and Δt out of temporary
memories and wait for the next event occurring on it.
Value of t1 will be compared to that of t2 so as to deter-
mine the next pair of coincidence events. Thus, the proc-
ess keeps on until the measurement finishes.
3. Analysis and Assessment of the Design
As physical side, test results for the block diagram in
Figure 1 are presented in Figure 3. The test measure-
ment was carried out with 60Co, activity ~20 kBq and
137Cs, activity ~100 kBq, measurement range of TAC
Figure 2. The block diagram of an “event by event” re-
cording digital coincidence spectrometer.
Figure 3. Test results as physical side for the configur ation in Fi gur e 1.

A Design Coinfiguration of an FPGA-Based Coincident Spectrometry System 161
was 500 ns. The obtained results showed that on the basis
of the time difference of pairs of coincidence events be-
tween the two detectors, the spectra corresponding to
60Co and 137Cs could properly be separated with high
precision. Drawback of the configuration was that the
performance of the system reduced because of fixed time
interval during the inputs of Start and Stop of TAC;
therefore, a modification of circuit design which can re-
place the role of traditional start and stop signals had to
be established. The modified circuit is called time-stamp
one. In the second configuration, the flexibility of vary-
ing timestamp at which a triggering event occurred will
allow us to overcome the aforementioned drawback. For
the semiconductor detector when the incoming radiation
interacts with the positions at the edge of crystal, the
charge collecting time will tend increasing. This leads the
rising edges of the pulses at the preamplifier’s outputs
pulled longer. If the operation frequency of the signal
processing circuit is not high enough, or algorithm for
determination of threshold-crossing time is not so good
(due to jitter or walk), the timing resolution of the system
will become worse significantly. If the frequency of the
circuit is called f0, error for determining time stamp will
be ±1/f0 (sec). For instance, if f0 is equal to 40 MHz, the
time error will be ±25 ns. To test this, PCI interfacing
unit has been programmed for running at 80 MHz, the
signals at the inputs (Start, Stop) of TAC were postponed
with a number of different time intervals owning to
nanosecond-delay module, Canberra. The final results
showed that error in determining the threshold crossing
time is 12.5 ns.
To overcome the above-mentioned restrictions as us-
ing HPGe detectors, it is better to add an algorithm capa-
ble of removing the pulses that have the slow rising
edges into DSPs. However, the additional algorithms can
make processing speed to become slow, especially for
determining the height of peak. Therefore, in this case
the third configuration should be used in order to over-
come this drawback (see Figure 4). This design allows
separating and processing properly timing signals. As a
result, the pulses having the slow rising edges are re-
Figure 4. The principal design of digital “event-event” coin-
cidence system with addition of filtering time circuit.
moved via filtering circuits before coming to the digital
processing parts will play a role in determining the
threshold crossing time. In the case of having no coinci-
dence signal, data reading algorithms have to be modi-
fied in order to remove amplitude value A1 or A2 because
the filtering circuit rejected almost of the non-suitable
pulses.
4. Conclusions
This work presents the two-different radiation measure-
ment systems for detecting pairs of coincidence events.
They are the traditional coincidence spectrometer and the
DSP-based one, respectively. Obvious drawback of the
first system is cumbersome in size, adjusting operation
and synchronizing signals among electronics stages. On
the contrary, both of the two digital systems can over-
come the above-mentioned drawback owing to the flexi-
bility of varying time-stamp window, and of the con-
figurations inside FPGA entities. In addition, the design
of system based on DSP techniques using FPGA allows
constructing a compact and impact coincidence spec-
trometer in which all of parameters are selected and con-
trolled by the application software. In the reality, DSP
algorithms are completely digitized by the Very high
speed integrated circuit VHDL, therefore, the digital
system is truly faster than the traditional system.
The significance and the importance of this work are
that it contributes partly into the application and devel-
opment of some radiation measurement systems based on
DSP techniques using FPGA. So far, the diagrams in
both of Figures 2 and 4 have been studying and fabric-
cating at the Department of Nuclear Physics and Elec-
tronics, NRI, Dalat. From the point of nuclear electronics
instruments, the design of system ought to be the basis
for development and application of coincidence meas-
urement techniques in the field of physics research and
radiation detection. Moreover, in the case of modifying
the structure of its hardware, it can also be used in con-
struction of a part of a Compton suppression spectrome-
try system—one of the most important gamma-ray spec-
trometers applied in experimental nuclear physics.
REFERENCES
[1] EG&G ORTEC Catalog, “Nuclear Instruments and
Systems Catalogue,” 1987.
[2] A. M. Hoogenboom, “A New Method in Gamma-Ray
Spectroscopy: A Two Crystal Scintillation Spectrometer
with Improved Resolution,” Nuclear Instruments, Vol. 3,
No. 2, 1958, pp. 57-68.
doi:10.1016/0369-643X(58)90092-6
[3] W. R. Leo, “Techniques for Nuclear and Particle Physics
Experiments,” Springer-Verlag, Berlin, Heidelberg, 1987.
[4] N. N. Dien, V. H. Tan and P. D. Khang, “The Gamma-

A Design Coinfiguration of an FPGA-Based Coincident Spectrometry System 162
Gamma Coincidence Spectrometer for Research on Nu-
clear Structure at DNRI,” Proceedings of International
Symposium on Instrumentation of Small and Medium Ac-
celerators, Tsukuba, 1996, pp. 128-131.
[5] P. D. Khang, N. X. Hai, V. H. Tan and N. N. Dien, “Ga-
mma-Gamma Coincidence Spectrometer Setup for Neu-
tron Activation Analysis and Nuclear Structure Studies,”
Nuclear Instruments and Methods in Physics Research,
Vol. 634, No. 1, 2011, pp. 47-51.
[6] A. Kimura, Y. Toh, et al., “Development of a Data Ac-
quisition System for a Multiple Gamma-Ray Detection
Method,” AIP Conference Proceedings, Vol. 769, Mel-
ville, New York, 2005.
[7] M. J. Koskelo, I. J. Koskelo and B. Sielaff, “Comparison
of Analog and Digital Signal Processing Systems Using
Pulse,” Nuclear Instruments and Methods in Physics Re-
search, Vol. 422, No. 1-3, 1999, pp. 373-378.
[8] J. M. Los Arcos, et al., “A New Digital Pulse Height
Analysis Method for Radiation Spectroscopy,” Nuclear
Instruments and Methods in Physics Research, Vol. 339,
No. 1-2, 1994, pp. 99-101.