dsp algorithms for fission fragment and prompt fission neutron spectroscopy
TRANSCRIPT
http://irmm.jrc.ec.europa.eu/
http://www.jrc.ec.europa.eu/
1
DSP algorithms for fission fragment and
prompt fission neutron spectroscopy
Sh. Zeynalov1,2, O.V. Zeynalova2, F.-J. Hambsch1, S. Oberstedt1, I. Fabry1
1)EC-JRC-Institute for Reference Materials and Measurements, Retieseweg 111, B-2440 Geel, Belgium2)JINR-Joint Institute for Nuclear Research, Dubna Moscow region, Russia
3
Digitized waveforms structure
Event = А1, А2, K, N
Cathode pulse used as a fission event
time reference
Energy released inside the 1st and the 2nd
half of the TGIC ND signal used for the neutron TOF
measurement after pulse shape analysis
A1
A2
K
N
The trigger pulse
L samples M samples
Waveforms sampled with 100
MHz WFD 12 bit amplitude
resolution
The actual trigger position L has the mean value (L-0.5)∆
and the dispersion σ =0.3 ∆
4
Analogue signal processing
Traditional analogue electronics approach when each module performs a dedicated signal processing.
Detector current is converted into a step pulse in a charge-sensitive preamplifier. The height of the step pulse conveys
information on the FF’s kinetic energy, released during deceleration in the working gas of the IC. The algorithm for precise
pulse height measurement is widely accepted for about 50 y in experimental nuclear physics and it is implemented in a
variety of commercially available electronic modules. In a spectroscopy amplifier the pulse undergoes first differentiation,
and then the result is integrated by a shaping filter in order to improve the SNR. The peak value of the shaped pulse is
measured with the help of an ADC and the numerical value is transferred to the PC, where the desired pulse height
distributions can be accumulated and displayed on demand.
5
DSP approach
Spectroscopic amplifier
Peak sensitive ADC
Multichannel analyser
CFD and TAC
Pulse shape analyser
And many other very useful devices….
The DSP approach uses a single waveform digitizer module and a variety of software to suit different experimental needs.
( ) (τ) ( τ) τ
0
V t I h t d∞
= −∫
0 0
( ∆) ( ∆) (( )∆)k n kn
n n
V k I n h k n V I h∞ ∞
= =
= − ⇒ =∑ ∑
Continuous and discrete form relations
between the detector current pulse (In) and the
preamplifier output signal (Vk) are as follows:
6
Basic maths for spectroscopy amplifier
0
( τ )( ), ( ) и (τ) τ ,
τ
In In O ut O ut Int In
k k k
dW kV V k V V k V V d
d
∞∆ −
= ∆ = ∆ = ∫
,1
Int Int InV V A V
k k k= × +
+.Out Int In
V V Vk k k
= −
0 0
(τ) ( τ)( ) ( τ) τ ( ) (0) (τ) τ
τ τ
t tInOut In IndV dW t
V t W t d V t W V dd d
−= − = −∫ ∫
τ 1 τ(τ ) exp , exp( )
τ
dWW
A d A A= − = × −
The differentiating formula
The kernel function describing a signal passing through an RC circuit
Sampling is the way to convert continuous signal to discrete
Matrix representation of discrete signals and equations
7
50000.00
0.01
0.02
0.03
0.04
0.05
Time [nsec]
Pu
lse h
eig
ht
[V]
RC = 800 nsec
Step function
CR-RC
RC2 - integration
RC3 - integration
RC4 - integration
RC5 - integration
4950
0
50
100
Cu
rre
nt
[arb
itra
ry]
Time [nsec]
100 2000
5000
10000
Co
un
ts
Pulse height [arb.]
100 2000
2000
4000
6000
8000
Co
un
ts
Drift time [arb.]
∑+
=
=ML
k
kIkT1
][*kknnnknk VhIIhV *)(* 1−=⇒=
Fission fragment pulse processing
8
FF angle determination
100 150 200
100
200
T90
T0(E)
Pulse height [arb.]
Drift tim
e [a
rb.]
1θ
2θ
0.0 0.5 1.00.0
0.5
1.0
CO
S(Q
2 )
COS(Q1)
FF 0 and 90for drift time the-T,T ,)cos(*))(()( 0o
09009090 ΘΕ−=− TTETT
city drift velo theis W distances;A -G theis d andG -C theis D where,*5.0
90W
dDT
+=
uesheight val pulse original and corrected are P P ,*
* OC
90
90 andTT
TPP
OC
σ+=
is the grid inefficiency factorσ
100 150 20040
60
80
100
120
140
T0(E
) [a
rb.]
FF kinetic energy [arb]
Backing side
Layer side
9
FF’s energy loss correction
80 90 100 110 120 130 140 150 160 1700
2
4
6
8
YIE
LD
[%
]
MASS [amu]
NPA617
Present measurement
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.050
75
100
125
150
175
Pu
lse h
eig
ht
[arb
.]
1/COS(ΘΘΘΘ)
Backing side
Layer side
100 150 2000.0
0.5
1.0
CO
S(Q
)
Pulse height [arb.]
50 1000
5000
10000
Co
un
ts
FF energy [MeV]
Backing side
Layer side
11
Event = А1, А2, K, N
Cathode pulse used as a fission event
time reference
Energy released inside the 1st and the 2nd
half of the TGIC ND signal used for the neutron TOF
measurement after pulse shape analysis
A1
A2
K
N
The trigger pulse
L samples M samples
Waveforms sampled with 100
MHz WFD 12 bit amplitude
resolution
Digitized waveforms structure
12
100 150 200 250
0.0
0.5
1.0
Pu
lse h
eig
ht
[arb
itra
ry]
Time [arbitrary]
Step pulse
Output of 4 order Butterworth filter
First the DSP signals are passed trough the 8th order
Butterworth filter in order to comply to the Nyquist criterion
The Butterworth filter of 2,4,6,8-th order
0.0 0.1 0.2 0.3 0.4 0.5
0.2
0.4
0.6
0.8
1.0
H( ΩΩ ΩΩ
)
Frequency
Betterworth 2
Betterworth 4
Betterworth 6
Betterworth 8
The band-limited continuous-time signal, with bandwidth of B Hz can be recovered from its samples provided
that the sampling speed Fs > 2B samples/sec. Thus, to assure a correct signal analysis the bandwidth of the signal
should be limited to 50 MHz.
Shannon sampling frequency
13
PFN time-of-flight spectroscopy
The left hand figure illustrates how the ND signals are transformed after passing the Butterworth filter. Properly
chosen cut-off frequency of the filter guarantees the same rise time for all ND pulses. The right hand figure illustrates
how the CFTM algorithm works with pulses having the same rise time, but different pulse heights. Signals are passing
the constant fraction of their pulse height at the same time instant. So the simplest digital realisation of the CFTM is to
get time instant at the constant fraction of the pulse height
1 00 15 0 200 2 50
0 .0
0 .2
0 .4
0 .6
0 .8
1 .0
Pu
lse
he
igh
t [a
rbit
rary
]
T im e [a rb itra ry]
1 .0
0 .5
0 .1 67
0 .2 692 , 0 .13 463 , 0 .04 488
2 4 0 2 6 0 2 8 0 3 00
0
5 0 0
1 0 0 0
1 5 0 0
Cu
rren
t [a
rbit
rary
]
T im e [n s e c ]
O r ig in a l N D p u ls e
N D p u ls e a fte r 8 o rd e r B u tte rw o rth f ilte r
14
Interpolation formulas:
))()((*)()( 1 kkkk tftftftf −∆+=∆++ Linear
32
2
33
12
333
12
3333
112
23
)2(*)2(*)2(*)(
))1()2((**)32()()(
2
))1(*2)2((*)()(2)(
)1(3)1(3)2(
)()(3)(3)(
)(*)(*)(*)(
+−+−+−=
+−+−+−−=
++−+−+−=
−−++−+
−+−=
+∆++∆++∆+=∆+
+
++
++
−++
kakbkctfd
kkabktftfc
kkkatftftfb
kkkk
tftftftfa
dtctbtatf
k
kk
kkk
kkkk
kkkk
ctbtatf kkk +∆++∆+=∆+ )(*)(*)( 2
2
1
2
**)(
)12(*)()(
2/))()(2)((
kakbtfc
katftfb
tftftfa
k
kk
kkk
−−=
+−−=
+−=
+
+ Parabola
Cubic parabola
)exp(*!3*
1),(
3
4τττ
τtt
tB −
= 4-th order Butterworth filter
15
300 400 500 600
1
10
100
1000
Co
un
ts
Time [0.5*nsec]
Prompt fission neutrons TOF distribution
Neutrons
Ph
oto
ns
Prompt fission neutron TOF distribution obtained using the developed CFTM
algorithm with cubic parabola interpolation. The flight path is 0.8 m and the FWHM
measured for the photon peak is 1.7 nsec
16
The pulse shape separationprinciple
100 200 300 400
0.01
0.1
1
10
100
Am
plitu
de
[a
rbit
rary
]
Time [arbitrary]
Neutron pulse
Photon pulse
Narrow window
Wide window
.
1 2
0 0
( ) , y ( ) , where ( ) - is i-th ND current waveform
T T
x I t dt I t dt I t= =∫ ∫
17
Neutron – Gamma separation
0
20
40
50
100
150
200
250
0
20
40
60
50 100 150 200 250
50
100
150
200
250
Neutrons
Photons
Illustration of the neutron – gamma pulse shape separation
principle.
18
Conclusions
• Digital signal processing (DSP) algorithms for FF and PFN spectroscopy have been
developed
• The algorithms are applied in an experiment with 252Cf(SF), using a fully digital
acquisition system with four 12bit/100 MHz WFD.
• DSP algorithms are developed as recursive procedures performing the signal
processing, similar to those available in various nuclear electronic modules such
as constant fraction discriminator (CFD), pulse shape discriminator (PSD), peak-
sensitive analogue-to-digital converter (pADC), pulse shaping amplifier (PSA).
• To measure the angle between FF and the cathode plane normal of the GTIC a new
algorithm is developed, having advantage over the traditional analogue pulse
processing schemes.
• Algorithms are tested by comparison of the results of the DSP data analysis for 252Cf(SF) with the data available from literature, demonstrating a superior quality of
the DSP technique over traditional analogue signal processing.