IEICE Electronics Express, Vol.6, No.

16, 11991204
Amplication of short
pulses in transmission lines
periodically loaded with
Schottky varactors
Koichi Narahara
a)
Graduate School of Science and Engineering, Yamagata University,
4316 Jonan, Yonezawa, Yamagata 9928510, Japan
a) narahara@yz.yamagata-u.ac.jp
Abstract: Pulse progagation on nonlinear transmission lines
(NLTLs), which are transmission lines with regularly spaced Schottky
varactors, is investigated for the amplication of short pulses. It is
found that the soliton developed in an NLTL experiences an exponen-
tial amplitude growth, when it couples with a co-existing voltage edge.
Keywords: solitons, nonlinear transmission lines (NLTLs), pulse am-
plication
Classication: Science and engineering for electronics
References
[1] R. Hirota and K. Suzuki, Studies on lattice solitons by using electrical
networks, J. Phys. Soc. Jpn., vol. 28, pp. 13661367, 1970.
[2] M. J. W. Rodwell, S. T. Allen, R. Y. Yu, M. G. Case, U. Bhattacharya, M.
Reddy, E. Carman, M. Kamegawa, Y. Konishi, J. Pusl, and R. Pullela,
Active and nonlinear wave propagation devices in ultrafast electronics
and optoelectronics, Proc. IEEE, vol. 82, pp. 10371059, 1994.
[3] Y. S. Kivshar and B. A. Malomed, Dynamics of solitons in nearly inte-
grable systems, Rev. Mod. Phys., vol. 61, no. 4, pp. 763915, 1989.
[4] T. Taniuti, Reductive perturbation method and far elds of wave equa-
tions, Prog. Theor. Phys. Suppl., vol. 55, pp. 135, 1974.
1 Introduction
It is well-known that a nonlinear transmission line (NLTL) dened as a
lumped transmission line containing a series inductor and shunt Schottky
varactor in each section succeeds in the development of solitons [1]. More-
over, the operation bandwidth of carefully designed Schottky varactors goes
beyond 100 GHz; therefore, they are employed in ultrafast electronic circuits
including the sub-picosecond electrical shock generator [2]. The resulting
short pulse can be applied to high-resolution measurement and high-speed
communication systems. An NLTL is useful for more than just short-pulse
generation. We found that it enables the amplication of short pulses. This
c
IEICE 2009
DOI: 10.1587/elex.6.1199
Received July 24, 2009
Accepted July 24, 2009
Published August 25, 2009
1199
IEICE Electronics Express, Vol.6, No.16, 11991204
article discusses the method of pulse amplication in an NLTL, together with
design criteria obtained analytically and several results of numerical evalua-
tions that validate the method.
2 Operating principle
Figure 1 shows the representation of an NLTL. L and C represent the series
inductor and shunt Schottky varactor of the unit cell, respectively.
n
and
I
n
show the line voltage and current at the nth cell. For later convenience, we
consider the case where each cell of the line is individually biased.
n
shows
the terminal voltage of the nth Schottky varactor. The capacitancevoltage
relationship of a Schottky varactor is generally given by
C(V ) =
C
0
_
1
V
V
J
_
m ,
(1)
where C
0
, V
J
and m are the optimizing parameter. Note that V < 0 for
reverse bias.
Figure 2 shows the operating principle of the amplication method. Fig-
ure 2 (a) shows the initial setup of signal application. The red short pulse
part, supported by a blue step-like part, is to be amplied. The top voltage
level is set to V
0
. The voltage level rst decreases up to V
1
, forming an
edge p
e
. The pulse to be amplied is set up after the voltage reaches V
1
.
First, the red pulse becomes solitonic due to the presence of varactors. Be-
cause the voltage levels of p
e
are greater than those forming the solitonic
pulse, the velocity of p
e
has to be smaller than that of the solitonic pulse, so
that the solitonic pulse overtakes p
e
(Fig. 2 (b)) and nally leaves p
e
behind
(Fig. 2 (c)). During this process, the solitonic pulse will be amplied due to
the eect of its coupling with p
e
, yielding the nal amplitude of the pulse to
be A
f
(>> A
i
) of Fig. 2 (c). Although the pulse propagation in Fig. 2 cannot
be characterized rigorously owing to the presence of varactors nonlinearity,
it is still possible to examine the eect of p
e
on the short pulse by using the
solitons perturbation theory [3], when we assume that the degree of voltage
variation of p
e
is much smaller than that of the short pulse.
In order to apply the perturbation theory, we rst derive the Korteweg-
de Vries (KdV) equation from the transmission equation of an NLTL with
varactors modeled by eq.(1), being followed by a modulation term that cor-
responds to the eect caused by the coupling with p
e
. We will see that the
Fig. 1. Equivalent representation of NLTLs. c
IEICE 2009
DOI: 10.1587/elex.6.1199
Received July 24, 2009
Accepted July 24, 2009
Published August 25, 2009
1200
IEICE Electronics Express, Vol.6, No.16, 11991204
Fig. 2. Operation principle of pulse amplication with an
NLTL. The temporal voltage waveforms at (a) the
input, (b) mid point and (c) output.
modulation term results in the exponential growth of the solitons amplitude
in eq.(18). This observation establishes our method.
The transmission equation of an NLTL is given by
L
dI
n1
dt
=
n1

n,
(2)
C(
n

n
)
d
n
dt
= I
n1
I
n.
(3)
When the pulse spreads over many cells, the discrete spatial coordinate n can
be replaced by a continuous one x, series-expanding
n1
up to the fourth
order of the cell length d as

n1

n

(x, t)
x
d +
d
2
2

2
(x, t)
x
2

d
3
6

3
(x, t)
x
3
+
d
4
24

4
(x, t)
x
4
. (4)
Applying this long-wavelength approximation to eqs.(2) and (3), we obtain
the evolution equation of the line voltage:
lc()

t
2
=

2

x
2
+
1
12

x
4
l
dc()
d
_

t
_
2
, (5)
where l and c are the line inductance and capacitance per unit length dened
as l = L/d and c = C/d, respectively. In order to derive the soliton equation
c
IEICE 2009
DOI: 10.1587/elex.6.1199
Received July 24, 2009
Accepted July 24, 2009
Published August 25, 2009
1201
IEICE Electronics Express, Vol.6, No.16, 11991204
from eq.(5), the voltage variables are series-expanded as
(x, t) =

i=1

(i)
(x, t), (6)
(x, t) = V
0
+

i=1

(i)
(x, t), (7)
for << 1. Note that V
0
> 0 for biasing Schottky varactors reversely.
Moreover, the following transformations are applied.
=
1/2
_
x
1

c
1
l
t
_
, (8)
=
3/2
t, (9)
where c
1
= c(V
0
). By evaluating eq.(5) for each order of , we can extract
the equation that describes the developing solitonic pulses [4]. It has been
shown that O() and O(
2
) terms give trivial identities, and O(
3
) terms
result in the following modulated KdV equation:

(1)

m
2

c
1
l(V
0
+V
J
)

(1)

(1)

+
1
24

c
1
l

(1)

3
+
m
2

c
1
l(V
0
+V
J
)

(1)

(1)

= 0
.
(10)
Scaling
(1)
, and as
(1)
= 18(V
0
+V
J
)/m, =

c
1
l

/9 and =

/6,
respectively, eq. (10) becomes

+

3

3
= R(

)
,
(11)
R(

) =
(1)

,
(12)
where = m/3(V
0
+V
J
).
For convenience, we briey review the fundamental properties of KdV
solitons. The standard KdV equation,

+
3

= 0, has the
one-soliton solution which is explicitly described as
= 2
2
sech
2
z, (13)
z = (

), (14)
= 4
2

, (15)
where is the parameter that corresponds to the solitons amplitude, which
is time-invariant for unmodulated ones.
In the framework of the solitons perturbation theory, the solitons ampli-
tude becomes time-dependent by the presence of R(

), and is described
by the following evolution equation of [3]:
d
d

=
1
4
_

dzR(z,

)sech
2
z
.
(16)
In order to investigate the situation shown in Fig. 2, we consider the case
where (x, t) is given by V
0
+
3/2
V
r
(x t/

c
1
l). Using eq.(8), it results in

(1)
= V
r

/6
.
(17)
c
IEICE 2009
DOI: 10.1587/elex.6.1199
Received July 24, 2009
Accepted July 24, 2009
Published August 25, 2009
1202
IEICE Electronics Express, Vol.6, No.16, 11991204
Originally, both the oor step-like and short pulses are applied to the NLTL,
while is set to identically zero. At this point, we think of the oor step-like
pulse as applied as . This corresponds to the situation where p
e
travels at
a speed of 1/

c
1
l with a spatial gradient of
3/2
V
r
.
Using eqs.(12), (13) and (17), the right-hand side of eq.(16) is calculated
be V
r
/18, thus giving
=
0
exp
_

V
r

18
_
,
(18)
where
0
corresponds to the unperturbed solitons amplitude. Because < 0,
grows if V
r
> 0. The gradient of p
e
shown in Fig. 2 satises this condition.
Moreover, the growth rate increases when the gradient of p
e
becomes larger;
therefore, the pulse gain in an NLTL can be controlled by the input waveform.
3 Numerical evaluation
We numerically solve eqs.(2) and (3) using standard nite-dierence time-
domain method for an NLTL with Schottky varactors having C
0
= 64.77 pF,
V
J
= 3.561 V and m = 1.259. L, V
0
, V
1
and the fall-time of p
e
are set to
100 H, 0.0 V, 1.0 V, and 30 s, respectively. Moreover, the total cell size
is 3000. The input pulse to be amplied has a form of
(t) = A
V
0
+V
J
m
sech
2
_
A
LC(V
1
)
_
1 +
A
6
_
t
_
, (19)
which represents the one-soliton waveform given by eq.(13) using the n and
t coordinates. For the present analyses, A is kept xed at 0.3.
The numerically obtained waveforms monitored at three distinct points
on an NLTL are shown in Fig. 3 (a), (b) and (c). The short pulse really
travels faster than the step-like one, and starts coupling with p
e
at 300 cell
distances from the input, and then completes the coupling with p
e
at 2100
cells from the input. As a result of the shock formation, the fall-time of
p
e
decreases as it propagates along the line. The oor step-like pulse has
suciently large voltage variation, so that its eect to the solitonic short
pulse cannot be treated perturbatively. However, the exponential growth of
the pulse amplitude is successfully demonstrated.
To see the growth of amplitude more clearly, the variations of the pulse
peak are shown in Fig. 3 (d). The black curve corresponds to the analysis of
the present loss-less NLTL, and the red one corresponds to the case where
the NLTL has 0.2 series resistance at each cell. The exponential increase in
the peak height is explicitly observed for both cases. From eq.(13), the pulse
width decreases as the amplitude increases. During the amplication process,
the pulse width is reduced so much that the discrete line structure inuences
the waveform, giving oscillatory temporal variations observed in Fig. 3 (d).
When the spatial pulse extent is reduced to a few cells, it will be relaxed by
decreasing the amplitude. Then the pulse is amplied again and its width
is reduced. This cycle explains the oscillations. Although it requires further
investigations to quantify how much resistance an NLTL allows, an NLTL
c
IEICE 2009
DOI: 10.1587/elex.6.1199
Received July 24, 2009
Accepted July 24, 2009
Published August 25, 2009
1203
IEICE Electronics Express, Vol.6, No.16, 11991204
Fig. 3. Numerical evaluations of pulse amplication using
an NLTL. The waveforms monitored at (a) 300,
(b) 1500 and (c) 2100 cell distant from input, and
(d) the temporal variations of the peak voltages.
The temporal range sandwiched by two red circles
in (a), (b) and (c) corresponds to p
e
of Fig. 2 (a).
succeeds in the pulse amplication even at the nite presence of electrode
resistance.
Note that the method does not need any sophisticated semiconductor pro-
cess technologies, because a unique nontrivial device is a Schottky varactor,
the simplest and fastest semiconductor device. Although the present analysis
deals with MHz pulses, we believe that it is equally possible to amplify very
short pulses with picosecond durations by using monolithically integrated
devices.
c
IEICE 2009
DOI: 10.1587/elex.6.1199
Received July 24, 2009
Accepted July 24, 2009
Published August 25, 2009
1204