Professional Documents
Culture Documents
Theory and Practice of Pulse Compression
Theory and Practice of Pulse Compression
Theory and Practice of Pulse Compression
CrossMark
View Export
Online Citation
AFFILIATIONS
Poly-Grames Research Center and Department of Electrical Engineering, Polytechnique Montreal, Montreal H3T 1J4, Quebec, Canada
a)
Author to whom correspondence should be addressed: muhibur.rahman@polymtl.ca
ABSTRACT
This paper presents a comprehensive analysis of pulse compression capability in hybrid and gyromagnetic non-linear transmission lines
(NLTLs). The corresponding theoretical analysis in the hybrid NLTLs is derived and discussed in detail with the generation and sharpening
aspects of leading and trailing pulse edges. The parameters responsible for pulse sharpening are examined and their corresponding pulse
compression capability is modelled by providing the output waveform while varying these parameters. A holistic overview and mathematical
I. INTRODUCTION NLTLs are usually used for pulse generation and edges
sharpening8,9 and they are categorized into non-linear electrical
Non-linear transmission lines (NLTLs) inherently possess the
capability to generate solitons having the ability to maintain their transmission lines (NLETL) and non-linear magnetic transmission
permanent pulse profile.1 Since non-linearity is essential for the lines (NLMTL).10 NLETL further consists of three different types
propagation of solitons, it cannot act in isolation. On the other including non-linear capacitive transmission lines (NLCTL), non-
linear inductive transmission lines (NLITL), and non-linear hybrid
hand, dispersion is a physical behavior that correspondingly coun-
transmission lines (NLHTL).10–12 It was previously explored that
teracts the balance of non-linearity in NLTL.2 Solitons that are
usually generated in a lumped NLTL fall within the category of NLETL and NLMTL have the ability of rising and fall time com-
KdV solitons, which are basically a single smooth hump.3 The KdV pression and the injected input pulse would be compressed after
solitons are usually based on the equation that approximately passing through NLTL. The minimum edge sharpening of a pulse
is limited by the dispersive properties of the deployed NLTL.13
describes NLTL and often can be written in the following form4:
A good approximation of rise and fall time reduction was given in
@u @u @ 3 u different scenarios14–16 of NLTL, and it is usually determined by
þ 6u þ 3 ¼ 0, (1) the difference among time delays of the corresponding sharpened
@t @z @z
edge of a pulse and peak of the waveform.
where the first term quantifies that it is a first-order evolution There exist another important class of NLTL that falls within
equation with respect to t, and the second and third terms account the NLMTL category, and it is termed as gyromagnetic NLTL.17
for non-linearity and dispersion, respectively. The soliton solution This type of NLTL comprises of continuous nondispersive NLTL,
for different cases was given in Refs. 5–7. in which the inner conductor is lined with ferrite rings and the
C0
C(V) ¼ pffiffiffiffiffiffiffi , (7)
jVj
L0
L(I) ¼ pffiffiffiffiffi : (8)
jIj
dIn
L þ RIn ¼ Vn Vnþ1 , (9)
dt
dVn
FIG. 2. Hybrid NLTL having (a) nonlinear capacitors configuration and (b) non- C þ GVn ¼ In1 In : (10)
dt
linear inductors configuration.
The forward traveling wave solution can be given as follows:
where I, V, and Q represent the transmission line current, voltage, LCω2 ¼ 2(1 cos k): (13)
and charges, respectively, while the conductance per unit length is
represented by G. T signifies the voltage of the nonlinear capaci- The wavelength that corresponds to the cutoff frequency ωc is
tance and is given as follows: given as follows:
1 2
T(Q) ¼ (Q0 þ ηQ3 ), (5) ωc ¼ pffiffiffiffiffiffi : (14)
C0 LC
where η represents the cubic nonlinearity strength and small-signal Now Eq. (13) can also be re-written as
transmission line capacitance is shown by C0. To show the shock
front’s steady-state form, Eqs. (2)–(5) can be solved as follows: LCω2
(1 cos k) ¼ ,
2
1 2 tan h(n x0 ) 2 (15)
Q (n) ¼ Qs 1
2
, (6) ω
2 w cos k ¼ 1 2 :
ωc
where n = x − vt, v and Qs represent the propagation velocity and
charge amplitude of the shockwave, w represents the width of the The line impedance can be determined from Eqs. (8)–(11) as
corresponding shock front, and x0 is a constant that describes the
shock front’s original position. As can be observed, the shock’s V0
Z ¼ Z 0 e jf ¼ , (16)
velocity increases as the shock’s amplitude and nonlinearity I0
strength increase. Also, increasing non-linearity decreases the width qffiffiffi
of the shock wave that correspondingly decreases the pulse where Z0 ¼ CL , sin f ¼ ωωc . In the case of a network constructed
duration. from T and pi sections,20 the matched terminating impedance can
Let us now take a look at the hybrid scenario depicted in be given by Eqs. (17) and (18), respectively,
Fig. 2, where the series inductance is based on current and the
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
shunt capacitance is dependent on voltage, resulting in a line with
ω2
a constant characteristic impedance that can be terminated without ZT ¼ Z0 1 2 , (17)
a variable characteristic impedance. Considering the case of the ωc
Z0 solitons while propagating along a line and ultimately its edge sharp-
ZΠ ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
: (18) ening will occur depending on the rise and fall time limit such that9
ω2
1 2
ωc π pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Trise, min ¼ L0 C(Vmax ), (26)
4
An approximate wave equation based on the combination of
Eqs. (9) and (10) is given as follows:
8:8 1
2 T f , min (10% 90%) ¼ rffiffiffiffiffiffiffiffiffiffiffiffi , (27)
d d d ωd (0) v
LC Vn þ R Qn þ LG Vn þ RGVn 1 1
dt dt dt f
¼ Vnþ1 2Vn þ Vn1 : (19)
where f = 0.4 V, Tf,min represents the fall time limit, and Trise,min repre-
sents the rise time limit. By varying an input rise time with constant
Expressing Vn+1 and Vn−1 in terms of following truncated
line length, we obtain 10%–90% data for the output pulse. It must
series:
also be noted that varying the voltage and current sharpens the rise
and fall edges until Bragg cutoff frequency. The shortest possible rise/
@ 1 @2
Vn+1 Vn + Vn þ V n fall edge is also termed as the shock edge limit. In order to validate
@n 2 @n2 the theoretical formulation, we compare the output response from
3
1 @ 1 @4 theory and simulation for ten section hybrid NLTL as shown in Fig. 3.
+ V n þ V n : (20)
6 @n3 24 @n4 It shows that the simulation and theoretical responses agree very well.
It must be noted that the designing of ten section NLTL is performed
Now, changing the coordinates with corresponding scaling in an advanced design system (ADS) and the simulation is extracted
and transformation as9 from it which will be explained in Sec. IV A.
The results in Eq. (22) show that a shock front propagating
x ¼ ε1/2 (n v0 t), along a nonlinear LC network will break into an array of solitons.
τ ¼ ε3/4 v0 t/24, (21) This will ultimately produce the modulation of the delivered power
and V ¼ εF0 u/2: to the load. Additionally, during this process, there will be an
1R G * ψ
ψ* þ ,ψ ¼ ,
¼
L C 24 (23)
ξ ¼ n υM t, and u ¼ 2V/F:
where
exists an optimum point where peak power is maximum. This III. THEORY AND DESIGN OF GYROMAGNETIC NLTL
effect is well modeled for hybrid NLTL as shown in Fig. 4(a), A continuous and nondispersive class of NLTL known as
where increasing RF load resistance increases pulse amplitude but gyromagnetic NLTL uses magnetic materials as the medium sepa-
at the cost of increased ringing. Increasing load resistance highly rating the line conductors. The construction of the gyromagnetic
increases the oscillation amplitude. This eventually leads to a
line is detailed in Fig. 5. As illustrated in Fig. 5, the magnetic com-
higher ringing, and it is modeled in Fig. 4(b), which shows the
ponents that are possibly ferrite rings are positioned side by side
trend between RF load resistance versus the figure of merit (FOM). along with the inner conductor.
FOM is defined using Eq. (28) as27 This kind of NLTL is advantageous over inductive NLTL due
to the capability of generating stronger pulse oscillations at higher
Pulse amplitude(V) frequencies. Additionally, it may generate a wide frequency spec-
FOM ¼ : (28) trum and use the stronger nonlinearity of the ferrite-based induc-
FWHM(ns) Ringing(%)
tance. However, this type of NLTL can only support rise time
compression.
We can derive the corresponding solitary wave solution and
pulse width estimation of a gyromagnetic NLTL by analyzing its
unit section. To this end, we consider a static axial bias in ferrite
such that
Hz ¼ Hbias : (29)
i
Hθ ¼ , (30)
πde
dm di
de ¼ , (31)
ln ddmi
di dv
¼ C0 , (32)
dz dt
dv d
¼ (Li), (33)
dz dt
dv df
¼ , (34)
dz dt
where f represents the circumferential flux per unit length. Then, sep-
arating the df/dt term into linear and magnetization-driven terms as
follows:
FIG. 7. A typical unit section of a nonlinear gyromagnetic transmission line.
df di (dm di ) dMθ
¼ L0 þ μ0 , (35)
dt dt 2 dt
hpffiffiffiffiffiffiffiffii pffiffiffiffiffiffiffiffi
where L0 signifies the saturated inductance slope per unit length and T ¼ ε1/2 t L0 C z , Z ¼ ε3/2 L0 C z/2, G ¼ ε1/2 G, I ¼ ε1/2 I,
dMθ /dt shows the corresponding rate-change of the circumferential
component of M. To obtain a numerical solution, the line is discre- (39)
tized into a series of axial elements having a length Δz. The corre-
sponding equivalent circuit of the discretized model is shown in where ε is a small parameter, and substituting Eq. (39) into Eq.
Fig. 6. dMθ /dt establishes the instantaneous magnitude of voltage (38), we get a modified KdV equation for the gyromagnetic NLTL
sources V1 Vm Vn for each section and can be given by case as
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3
(dm di ) dMθ β 1 C0
V m ¼ μ0 Δz: (36) 6 þ 7
2 dt 6 6L0 C 12 C 7
u ¼ Is sech6 7
FIG. 9. Circuit model of the designed hybrid NLTL with capacitive and inductive FIG. 10. Varying the number of sections (n) of the designed hybrid NLTL and
losses. its corresponding effect on output pulse.
Parameters Remarks
Increasing TR is set to increase the
Rise time of input pulse, TR
number of oscillations as well as
increase the output pulse duration.
Amplitude of input pulse If amplitude is too high, oscillations
will shift upward and produce
distortions.
Number of NLTL sections, n Increasing n,
(a) increases the amplitude of
output peak voltage;
(b) considerable reduction in
ringing; and
(c) decreases pulse duration and so
FWHM of the output pulse.
Resistive load, RL Increasing RL,
(a) increases the amplitude of
oscillation in the output voltage;
FIG. 11. Varying load resistance (RL) of the designed hybrid NLTL and its (b) increases the amplitude of
corresponding effect on output pulse. ringing;
(c) increases pulse duration; and
(d) there exists RL where the
is seen to remain constant as the rise time varies until the line is optimum power is maximum and,
able to produce solitary waves. Although the pulse duration is cons- thus, has the best output
tant, the section of the flat top decreases as the rise time increases, performance.
thus limiting the number of cycles for the same frequency. As a Capacitive nonlinearity Steep capacitive nonlinearity:
magnetic flux through the ferrite rings. The created model is built
on the work suggested by Dolan,21 in which the effects of magneti-
zation of the ferrite are stated using the one-dimensional (1D)
form of the LLG equation and represented by voltage sources as
nonlinear elements. The proposed model’s schematic is already
shown and discussed in Sec. III as Fig. 6.
For the analysis of pulse edge sharpening and pulse compres-
sion behavior, we modeled a lossless gyromagnetic NLTL driven by
a Gaussian-shaped pulse having an amplitude of 5 V. The generator
resistance and load in the circuit model are also 50 Ω, matching the
line characteristic impedance. The relative saturated permeability is
5; the relative permeability and permittivity of the dielectric are 2.1
and 1.0, respectively; the ferrite relative permittivity is 16; the
ferrite outer diameter is 1.4 mm; the diameter of the inner conduc-
tor is 0.8 mm; and the diameter of the outer dielectric is 2.5 mm,
having the optimum selected number of sections n is 12.
Figure 13 shows that the corresponding output pulse is sharp-
ening by changing the number of sections (n) of the designed gyro-
magnetic NLTL. However, if we increase the number of sections,
the simulation running time is also going to increase considerably.
The impact of the damping parameter (α) on rise time compres- FIG. 14. Variation of rise time with damping parameter of the gyromagnetic
sion is demonstrated in Fig. 14. The dimensionless damping NLTL.
parameter, whose corresponding value for this simulation is
spanned between 0.01 and 1, is also investigated in detail.
According to Fig. 14, increasing the damping parameter results in a previous literature that a higher input voltage shortens the rise
consistent decrease in the output pulse rise time. Additionally, we time. For instance, in Ref. 22, the 3D form of the LLG equation23
can see in Fig. 14 that the output waveform variation is very small was used, while in Refs. 24 and 25, the 1D form of the respective
FIG. 13. Varying the number of sections (n) of the designed gyromagnetic FIG. 15. Variation of rise time with an input voltage amplitude of the designed
NLTL and its corresponding effect on output pulse. gyromagnetic NLTL.
FIG. 17. Block diagram of the measurement setup along with prototype (biasing
wires are not shown in the prototype).
FIG. 16. Variation of the rise time with saturation magnetization of the designed performance. Based on the above discussion, it is clear that gyro-
gyromagnetic NLTL. magnetic NLTL can perform rise time compression and this com-
pression is superior in comparison to inductive NLTL. This is due
to the reason that inductive NLTL utilizes only inductive nonlinear-
Conclusively, it is observed that we have maximum FOM based on ity of non-linear inductors while gyromagnetic NLTL utilizes the
Eq. (28) having n = 12, α = 0.5, and saturation flux density of stronger nonlinearity of the ferrite-based inductance.
0.275 T in the case of gyromagnetic NLTL. These parameters are The overall summary of the parametric analysis of the gyro-
helpful in designing gyromagnetic NLTL having our desired pulse magnetic NLTL is tabulated in Table II to see the behavior of dif-
VII. CONCLUSIONS
A comprehensive pulse compression capability analysis of
hybrid and gyromagnetic NLTLs is presented and discussed. Both
classes of NLTLs are theoretically analyzed and their corresponding
lumped-element models are developed. Parameters principally
responsible for pulse sharpening in hybrid NLTLs include the
FIG. 20. Comparison of output response from theory, simulation, and measure- number of NLTL sections, resistive load, input pulse variation,
ments of ten section hybrid NLTL. capacitive, and inductive nonlinearity. It is shown that there exists
an optimum value for each parameter selection where the output