Copyright © 2016 American Scientific Publishers Journal of

All rights reserved Computational and Theoretical Nanoscience

Printed in the United States of America Vol. 13, 4510–4514, 2016
RESEARCH ARTICLE

Research of Single Pulse Trapezoidal Shaping

Algorithm in Digital Nuclear Spectrum Measurement
Min Wang, Jianbin Zhou∗ , Xu Hong, and Ying Liu
College of Nuclear Technology and Automation Engineering, Chengdu University of Technology, Chengdu 610059, China

In order to improve the performance of digital spectrum measuring instrument, efficient recursive
algorithm is used to achieve fast digital pulse shaping, and real-time trapezoidal pulse instead of
the traditional analog pulse, so that the subsequent amplitude extraction and peak amplitude accu-
mulation processing become easier. According to the experimental verification, realization of this
technique improves the performance of spectrum measuring instrument. Especially the single pulse
trapezoidal shaping technology make the energy resolution of the spectrum measuring instrument
significantly improved.
Keywords: Recursive Algorithm, Single Pulse Trapezoidal Shaping, Digital, Energy Resolution.

1. INTRODUCTION The second is based on discrete components designed to

Digital nuclear measuring instrument is the development achieve S-K Gaussian forming filter, which the module
trend of the new generation of nuclear instruments. Com- function is single and not convenient for system upgrade.
pared with analog nuclear measuring IP: [Link]
instrument, it On:
has aMon, 28Digital
Oct 2019 [Link]
Gaussian shaping compared to trapezoidal shap-
Copyright: American
better core stability, noise immunity and fidelity. The study Scientific Publishers
ing, the
Delivered by Ingenta falling edge more slowly, shaping longer and not
of nuclear signal processing method is an important con- suitable for the occasion of high count rate.
tent of digital nuclear measurement instrument. In order to Based on the comparative analysis, this study chose
make the digital nuclear instrument can have better energy the trapezoidal pulse shaping methods. Trapezoidal pulse
resolution; the nuclear signal must be appropriately filter shaping method is Z transforms and function convolution.
forming. Using the Z transform to achieve trapezoidal shaping algo-
The research on digital processing of nuclear mea- rithm is complex, and Z transform is difficult to imple-
surement instrument has nearly 40 years of history. ment on the hardware system. The function convolution
In the 1970s, Koeman et al. had used digital methods at algorithm can be written in the form of recursive in time
nuclear spectrum measurement signal pulse sampling and domain and convenient for hardware implementation.
filtering;1 Jordanov et al. introduced the digitized pulse This paper describes single pulse shaping algorithm and
shaping into nuclear measuring instruments, and success- continuous pulse shaping algorithm, which based on func-
ful implementation of the trapezoid forming at 1994.2 tion convolution method to achieve the trapezoidal shap-
In this study, based on the above, the shaping algorithm
ing. Through further study and experiment, single pulse
was further improved and optimized.
shaping is more effective to avoid the accumulation of
Pulse shaping method is divided into two categories,
errors; the energy resolution of the nuclear energy spec-
analog and digital. Among them, the digital pulse shap-
trum measurement instrument was further improved.
ing method is divided into trapezoidal (triangular) and
3
Gaussian shape. The existing methods have the following
deficiencies:4–11 Analog circuit pulse shaping technology 2. MATHEMATICAL DESCRIPTION OF
will increase the system hardware design, system stability TRAPEZOIDAL PULSE
is poor and system upgrade is difficult. Digital pulse shap- Sharp pulse is the most ideal pulse in high resolution
ing has two implementations of Gaussian shaping. One is digital nuclear energy spectrum measurement instrument,
the wavelet analysis technique, which the great amount of because it can achieve the best signal-to-noise ratio, but it
calculation and not suitable for hardware implementation. is difficult to obtain in practice.
Nuclear detector signal is the exponential decay signal,
∗
Author to whom correspondence should be addressed. which superimposes on some noise components,6 as shown

4510 J. Comput. Theor. Nanosci. 2016, Vol. 13, No. 7 1546-1955/2016/13/4510/005 doi:10.1166/jctn.2016.5312
Wang et al. Research of Single Pulse Trapezoidal Shaping Algorithm in Digital Nuclear Spectrum Measurement

is used to realize the trapezoidal shape of recursion algo-

D rithm. The algorithm analysis in continuous time domain,
A

RESEARCH ARTICLE
and then apply the results to the discrete time domain.
Assuming the output function st is a linear time-
Vo

y1 y4 invariant systems, is a given convolution integral:

 +
st = vt  ht − t   dt  (4)
−

t Where, vt is the input signal, ht is the impulse

tb
y
response of the system. The real-time signal processing
y2 system must be causal; the output of fixed time is relying
on the present and previous input signal. Therefore, the
integral upper limit of Eq. (4) uses t to evaluate the output
Fig. 1. Diagram of trapezoidal shape. system. Its purpose is to draw a causal system with finite
impulse response, convolution function must be simple and
in the mathematical expressions of formula (1). easy to implement. In this study, using Eq. (4) to make the
input exponential signal converted into trapezoidal signal.
Vi = Ae−t/ ut (1)
(1) The trapezoidal response in continuous time domain
Where:  is the time constant, ut represents step func-  Convolution of negative exponential signal and gate
tion, A is the amplitude of the signal pulse. function.
The so-called trapezoidal shape is the exponentially First, the average mobile response system is constructed
decaying pulse forming an isosceles trapezoid pulse, trape- by using a rectangular cell gate function convolved with
zoidal bevel edge and top width can be adjusted, when the the input exponential signal. The process can be described
top width taken zero, trapezoid becomes a triangle. in Eqs. (5), (6):
As shown in Figure 1,7 11 the rising edge width of the t 
trapezoidal shaping is ta , the total width of the waveform P t = et −t/ dt  = 1 − e−t/  0 ≤ t ≤ T2 (5)
0
is tc , tc = ta + tb , A is the amplitude of the pulse signal,  T2 [Link]
IP: [Link] On: Mon, 28 Oct 2019 
ut is a step function: et −t/ dt  = e−t/ eT2 / − 1 t > T2 (6)
P t = Publishers
Copyright: American Scientific
⎧ Delivered by Ingenta 0
⎨1 t > 0
ut = Note: when t < 0, the response of the system is 0.
⎩  Convolution of negative exponential signal and saw
0 t < 0
tooth function.
According to the function expression, the expression of Another simple convolution function is convolved with
trapezoid is: saw tooth function of time is T . The response of the sys-
4
tem can be expressed as follows:
Vo = yi t (2)
i=1
t
t  −t/
And: ⎧ rt = te dt  = t −  2 1 − e−t/ 
0
⎪
⎪ Atut
⎪
⎪ y1 = 0 ≤ t ≤ T1 (7)
⎪
⎪ ta
⎪
⎪ 
⎪
⎪ T 1
t  −t/
⎪
⎪ At − ta ut − ta  rt = te dt  = e−t/  + eT1 / T1 − 
⎪
⎪ y = −
⎨ 2 ta 0

(3) t > T1 (8)

⎪
⎪ At − tb ut − tb 
⎪
⎪
⎪ y3 = −
⎪ When t < 0, the response of the system is 0.
⎪
⎪ ta
⎪
⎪ 
⎪
⎪ At − tc ut − tc 
Synthesis of the trapezoidal pulse.
⎪
⎪
⎩ y4 = The following discussion synthesis conditions of trape-
ta zoidal pulse. According to the function defined above, we
In Figure 1, D is the width of the trapezoid flat top, can define f t as:
tb = ta + D.
f t = rt + pt + apt − T1  (9)

3. REALIZATION OF THE Where a is parameter. For simplicity, the output sig-

TRAPEZOIDAL SHAPING nal can be considered a function of the time interval is
Convolution signal is the signal and system analysis essen- fixed. The start and end with a symbol to represent inter-
tial mathematical methods. Function convolution method val, assuming T1 ≤ T2 , for example, f01 means f t while

J. Comput. Theor. Nanosci. 13, 4510–4514, 2016 4511

 Research of Single Pulse Trapezoidal Shaping Algorithm in Digital Nuclear Spectrum Measurement Wang et al.

T1 ≤ t ≤ T2 . Use this way to define the time T3 = T2 + T1 , Where vn is the amplitude of n, l is the length of the
Eq. (9) can be written as the following three independent convolution function, vn − l is the amplitude of n which
equations.
RESEARCH ARTICLE

interval l. Since the offset is 0, the initial condition of the

offset of the output signal can be defined as:
f01 t =  (10)
f12 t =  2 1 − e−t/  + e−t/  + eT1 / T1 −  vn = 0 n<0 (19)
+ a1 − e−t−T1/  The second step is to define the recursive convolution
=  + a + e
2 −t−T1 /
T1 −  − a = T1  (11) algorithm of saw tooth function. Assuming that slope of
the bevel edge is the same; the form of recursive convolu-
f23 t =  2 e−t/ eT2 / − 1 + e−t/  + eT1 / T1 −  tion can be written as:
+ a1 − e−t−T1/ 
rn = rn − 1 + pn − vn − kk n≥0 (20)
=  2 e−t−T2 / + T1 −  (12)
Where pn represents the average mobile length of k.
Trapezoid is symmetrical, so the choice of a = T1 − . From the previous discussion, the trapezoidal respond
When T3 = T2 + T1 , f23 t signal attenuation to 0. There-
in continuous time domain can be written in a recursive
fore, the system response can be written as follows:
algorithm form in discrete time domain.
st = f t + qt (13) Suppose M is delay time constant of the exponential
signal, k is trapezoidal rise time, the delay time of the
In order to make st be a symmetric function, qt can
trapezoid flat top m = l − k. Response system is obtained
be defined as:
from Eq. (17) can be written as:
When − < t < T2 , qt = 0
When T2 < t < T3 ,
sn = rn + Mpn + k − Mpn − k − rn − l (21)
q23 t = −t − T2  −  2 1 − e−t−T2 /  (14)
According to the Eqs. (18) and (20),
When t > T3 , Set
q3 t = − 2 e−t−T 2 /

IP: [Link] (15)Mon, 28 k
On: Oct 2019 [Link]
Copyright: American Scientific = vj − vj − k − vj − l + vj − k − l (22)
d l jPublishers
According to Eqs. (7), (8), (14), (15), when T2 is atten-
Delivered by Ingenta
uation, qt and rt has the same waveform, but of oppo- The input signal of the conditional response system can
site polarity. be written as the recursive form:
qt = −rt − T2  (16)
Therefore, the characteristic of the system is nega- sn = sn − 1 + pn + d k l nM n ≥ 0 (23)
tive exponential pulse turn into trapezoidal pulse. When
T1 = T2 , the trapezoid becomes a triangle. Here:
According to the distribution properties of the convo-
lution function, impulse response of the system can be p n = p n − 1 + d k l n n ≥ 0 (24)
written as:
Equations (22)–(24) is the recursive algorithm model of
ht = h1 t+h2 t+T1 −h2 t −T1 −h1 t −t2  (17) trapezoidal, the algorithm can divide the input exponential
signal generates a recursive trapezoidal. Where k is the
Where h1 t and h2 t is the impulse response of the
trapezoid bevel length, l is the bevel length plus upper
hypotenuse and the baseline system, and T1 =  in this
flat, M is the input pulse shaping time. When k = l, it is
formulas.
(2) recursive form of trapezoidal pulse in the discrete a triangle, the output amplitude is normalized using the
time domain following formula.
The input exponential signal is obtained by cycle col-
V _out n = Sn/l/M (25)
lection. Therefore, the measurement time is divided into
units of measurement cycles, input signal i can be written According to (23)–(25), we could use fewer hardware
as vi. Signal convolution is divided into several steps.
modules to form a simple pulse processing circuit.
The first step is the synthesis of a recursive algorithm
which is convolution of exponential signal and gate func-
tion. Exponential decay signal convolution with rectangu- 4. IMPROVEMENT OF TRAPEZOIDAL
lar function signal is equivalent to average mobile, it’s SHAPING ALGORITHM
recursive form can be written as follows:
Experiments show that the error accumulation of con-
pn = pn − 1 + vn − vn − l n≥0 (18) volution algorithm to produce certain error pulse after

4512 J. Comput. Theor. Nanosci. 13, 4510–4514, 2016

Wang et al. Research of Single Pulse Trapezoidal Shaping Algorithm in Digital Nuclear Spectrum Measurement

continuous pulse shaping. And, compared with the analog 160

forming, energy resolution is not improved much. Such 140
120

RESEARCH ARTICLE
as measuring characteristic X-ray of the sample, the use

Voltage(mV)
100
of continuous pulse shaping methods obtains the energy
80
resolution is just about 180 eV.
60
Based on this, this study proposes using single pulse 40
shaping instead of continuous pulse shaping. The so- 20
called single pulse shaping refers to the process of pulse 0
shaping, when a complete pulse shaping after the com- –20
0 200 400 600 800 1000 1200
pletion of the recursive accumulated value of zero and
then forming the next pulse. Using single pulse shap- Time(uS)
ing method to deal with the same sample, the energy Fig. 4. Pulse signal after simulation Gaussian shaping.
resolution is about 165 eV, which has improved signif-
icantly. Experiments demonstrate the superiority of this
Table I. Comparison of different forming modes test of 55
Fe (detector
method. nominal resolution is 156 eV).

Forming mode Energy resolution/eV

5. THE EXAMPLES AND DISCUSSION OF
THE TRAPEZOIDAL SHAPING Gaussian shaping 180
Trapezoidal shaping 165
In this paper, construction of the experimental system: Triangular shaping 165
choose Si-PIN detector, the use of X-Tube excite the sam-
ple, using 55 Fe standard radioactive sources, signal collec-
tion selects 20 M high speed ADC, multi-channel pulse capability, less affected by the environment, and the system
processing adopts FPGA as main control chip, which can upgrade is very convenient. The key technologies to imple-
control ADC high-speed synchronous implementation pro- ment the system in the following two parts:
cess of collecting data cache, pulse filter, pulse shaping, 1. Choose the appropriate AD chip, the collected digital
amplitude analysis, spectrum count, etc. nuclear pulse signal;
The system processing speed isIP:fast, real-time perfor-
[Link] On: Mon,2.28FPGA
Oct 2019realization
[Link]of the established mathematical
mance, strong flexibility, stability, strong anti-jamming
Copyright: American Scientific
model. Publishers
Delivered by Ingenta
Many factors affecting energy resolution of nuclear
350 measuring instrument, therefore, the energy resolution of
300 the experimental test conditions for all tests are consistent.
250 As shown in Figures 2–4 is using the different sig-
Voltage(mV)

200 nals obtained from the experiment system for actual test-
150 ing, in which Figure 2 is measured using a trapezoidal
100 shaped pulse signal; Figure 3 is measured using a trian-
50 gular shaped pulse signal; Figure 4 is the use of simula-
0 tion Gaussian shaped pulse signal. As can be seen from
–50 these three figures, trapezoidal shape fall time shorter than
0 200 400 600 800 1000 1200
Gaussian shape, which more suitable for high count rate.
Time(uS)
In these three different forming modes, the energy res-
Fig. 2. Pulse signal after trapezoidal shaping. olution of the instrument is listed in Table I.
Nuclear measuring instruments using single pulse trape-
1000
zoidal shaping (triangular shaping) compared with the use
of Gaussian shaping, the energy resolution has greatly
800 improved.
Voltage(mV)

600

400 6. CONCLUSIONS
This study introduces the function of convolution method
200
is used to realize the trapezoidal shaping, establishes the
0 mathematical model of trapezoidal shaping and recur-
sive algorithm in time domain. Through simulation, sum-
0 200 400 600 800 1000 1200
marizes the practical problems encountered in shaping.
Time(uS)
Aiming at these problems, the algorithm model is opti-
Fig. 3. Pulse signal after triangular shaping. mized to improve, and the single pulse trapezoidal shaping

J. Comput. Theor. Nanosci. 13, 4510–4514, 2016 4513

 Research of Single Pulse Trapezoidal Shaping Algorithm in Digital Nuclear Spectrum Measurement Wang et al.

algorithm is proposed. The improved shaping algorithm References

compared with other shaping method has the following 1. H. Koeman, Filtering of Signals Obtained from Semi-Conductor
Radiation Detectors, Katholieke Universiteit Nijmegen, The
RESEARCH ARTICLE

advantages:
Netherlands (1973).
First, compared with the analog forming, it will reduce 2. V. T. Jordanov and G. F. Knoll, Nucl. Instrum. Meth. A 345, 337
hardware costs, improve the system stability and ease of (1994).
system maintenance and upgrade. 3. H. Koeman, Nuclear Instruments and Methods 123, 181 (1975).
Second, compared with a Gaussian shaping, the forming 4. F. Hilsenrath, H. D. Voss, and J. C. Bakke, IEEE Transactions on
Nuclear Seience 32, 145 (1984).
time of trapezoidal shaping is shorter. Trapezoidal shaping
5. R. E. Chrien and R. J. Sutter, Nuclear Instruments and methods in
improve the system count rate at the same time ensure that Physics Research, Section A: Accelerators, Spectrometers, Detectors
energy resolution do not decline. A 249, 421 (1986).
Third, the width of the trapezoidal shaping can be real- 6. J. Puncochar, Transactions of the VSB-Technical University Ostrava
time adjustment, which matches the input pulse form- 5, 232 (1995).
7. Y. Zhao, Y. Cai, and X. Yang, Thermal Science 3, 20 (2016).
ing time. It suit different measurement occasions when
8. V. T. Jordanov and G. F. Knoll, IEEE Transactions on Nuclear Sci-
equipped with different detectors, to achieve the require- ence 42, 683 (1995).
ment of the high count rate or high energy resolution. 9. Y. Zhao, Y. Cai, and D. Cheng, Multimedia Tools and Applications
Fourth, the single pulse shaping method is proposed for 19, 33 (2016).
the first time, and it is a further research on the existing 10. T. Valentin and G. F. Knoll, Nuclear Instruments and Methods in
Physics Research, Section A: Accelerators, Spectrometers, Detectors
convolution method. It avoids the influence of cumulative and Associated Equipment A345, 337 (1994).
error, which does a help to improve the energy resolution 11. S. G. Chen, S. Y. Ji, and W. S. Liu, Acta Phys. Sin.-.Ch. Ed. 57, 2882
of nuclear measuring instruments. (2008).

Received: 19 November 2015. Accepted: 28 November 2015.

IP: [Link] On: Mon, 28 Oct 2019 [Link]

Copyright: American Scientific Publishers
Delivered by Ingenta

4514 J. Comput. Theor. Nanosci. 13, 4510–4514, 2016