You are on page 1of 20

MODELING AND ANALYSIS

OF DIFFERENTIAL VCOs



A. BUONOMO and A. LO SCHIAVO

Dipartimento di Ingegneria dellInformazione,
Seconda Universit degli Studi di Napoli,
Via Roma, 29 - 81031 Aversa (CE), Italy.








Index terms: nonlinear circuits, perturbation methods, asymptotic methods, oscillators,
RF circuits, voltage-controlled oscillators.





Correspondence to:
A. Buonomo, Dipartimento di Ingegneria dellInformazione, Seconda Universit degli
Studi di Napoli, Via Roma 29, I-81031 Aversa (CE), Italy. e-mail:
antonio.buonomo@unina2.it . Telephone: +39 081 5010244, fax: +39 081 5037042





_________________________________________________
This paper was supported by the Italian Ministero dellUniversit e della Ricerca
Scientifica e Tecnologica.


1
ABSTRACT
We perform a systematic nonlinear analysis of differential Voltage Controlled
Oscillators (VCOs). An explicit nonlinear differential equation for them is
derived in a perturbation form, using standard models for the devices, either
bipolar or MOS. A solution method is then presented for a general equation,
valid for all possible second-order oscillators, which enables transient and
steady-state solution to be obtained providing explicit expressions for the
transient start-up, steady-state amplitude, harmonic content and frequency
deviation of the oscillator output waveform. The expressions enable us to analyze
and synthesize all second-order oscillators and may be useful in phase noise
evaluation.

I. INTRODUCTION
Technical problems posed by the realization of fully integrated LC-oscillators,
suitable for radio-frequency (RF) applications, have triggered a renewed interest
for their study. This interest is mostly aimed at obtaining a deeper insight into
mechanisms of phase noise in Voltage Controlled Oscillators (VCOs), whose
performances critically affect the stringent specifications of modern wireless
systems [1]-[4]. Accurate modeling of VCOs phase noise, needed to optimize
their design, requires a sound nonlinear theory, to which many relevant
contributions have been recently devoted [3],[4]. On the other hand, it is often
necessary to evaluate the VCO transient behavior too, for which a
computationally inexpensive tool is desirable [5],[6].
The present paper is devoted to the development of the nonlinear analysis of
the commonly used differential VCOs. Using the standard models of bipolar and
MOS devices, we derive an explicit nonlinear differential equation of VCOs,
either bipolar or MOS, which is arranged in a form involving a perturbation
parameter [7],[8]. These circuits are analyzed in the framework of the wide class
of the so-called negative differential-resistance oscillators [9],[10], of which the
differential VCOs are particular examples. To this end, we consider the most
general perturbation equation of second-order oscillators and present a solution
2
method of it, enabling transient and steady-state solution to be obtained
providing explicit expressions for the transient start-up, steady-state amplitude,
harmonic content and frequency deviation of the oscillator output waveform.
Although the oscillation can be rigorously predicted with the desired accuracy,
we focus on the expressions for the first approximations only, which are accurate
enough for practical purposes. The expressions enables us to analyze and
synthesize all second-order oscillators and may be useful in phase noise
evaluation.


II. PERTURBATION MODEL OF THE DIFFERENTIAL VCOS
The present Section is devoted to the derivation of the describing equations of
the differential VCOs, either bipolar or CMOS. Firstly, we consider the well-
known tail-current biased differential LC oscillator shown in Figure 1. The
behavior of this circuit can be efficiently described considering the nonlinear
characteristic ( ) v i around the quiescent operating point, 2
0 0 2 1
I i i i
E E E
,
of the active part of the oscillator, which is made up of the composite one-port
formed by the two cross-coupled transistors and the current source
0
I . This
characteristic can be analytically calculated using the Ebers-Moll model for the
transistors and observing that v v v
BE BE
+
1 2
, v v
BC

1
and v v
BC

2
.
Imposing that
2 1 0 E E
i i I + , we find

( )
( )
( )

,
`

.
|
+
+ +
+
+

T
T T
T
V v
V v V v
F F
V v
s
F
T BE
e
e e
e I
I
V v
1
2 2
1
ln
0
1

. (1)

Then, accounting for that 2
0 2 1
I i i i
B C
+ , the following expression can be
easily found for the nonlinear characteristic
3

( ) ( )
( ) ( )
]
]
]

,
`

.
|
+
+
+
+ +

,
`

.
|
+
+
+

T T T
T
T T T
T
V v
F
F
V v V v
F
V v
s
V v V v
R
s V v
F
V v
F
e e e
e
I
e e
I
e
e
I I
v i
1
1
3 4
1
1
1
1
2
2
0 0

(2)

whose derivative at the origin is

,
`

.
|
+ +
,
`

.
|


R F T
s F
T
V
I
V
I
v i g

2
2
2
1
4
1
2
1
) ( '
0
. (3)

The differential oscillator under investigation is thus essentially equivalent to the
second-order oscillator shown in Figure 1 (b), which is composed of a locally
active one-port connected to an LC-tank whose losses are attributed to a parallel
resistance R. Using the differentiation operator D with respect to the time t ,
the circuit equation is written in the form
( ) v i D
C
v
C L
D
C R
D
1 1 1
2

,
`

.
|
+ + (4)
and can be formulated as a regular perturbation problem [7],[8] by introducing
the perturbation parameter as the inverse of the circuit quality factor Q, that is

C R Q
0
1 1

(5)
where C L 1
2
0
. Using the dimensionless variables t t
0
and
T
V v x ,
equation (4) takes the form
( ) ( ) x n D k x D D 1
2
+ + (6)

where D is the derivative with respect to the time t ,
T
V I R k
0
and

4

( ) ( )
( )
( ) ( ) . 1
1
3 4
1
1
1
1
2
1
2
0
0
]
]
]

,
`

.
|
+
+
+
+

,
`

.
|
+
+
+

x
F
F
x x
F
x
s
x x
R
s x
F
x
F
e e e
I e
I
e e
I
I
e
e
x n

(7)

The driving-point characteristic (2) of the nonlinear one-port, which is shown in
Figure 2, can be approximated through the more simple function

( )
T
V v
e
I I
v i
+
+
1

2
0 0
(8)
whereby the differential conductance at the origin is calculated as

2
) ( '
m
g
v i g
2

0
T
m
V
I
g (9)
and the nonlinearity ( ) x n becomes

x
e
x n
+
+
1
1
2
1
) ( . (10)
In this case, equation (6) can be written in the form
( ) x f D k x D
Rg
D
m
1
2
1
2

]
]
]

+
,
`

.
|
+ (11)

with ( ) ( ) 4 x x n x f + . It is easy to realize that this equation is obtained from the
circuit in Figure 1(b) extracting from the one-port its linear part and representing
it as a negative conductance 2
m
g in parallel to the tank. Then, let R g
C
2
be the value of
m
g which compensates the losses of the LC-tank and
r
g the
remaining part of
m
g , i.e.
C m r
g g g , equation (11) is reduced to

( ) x f D k x D
g R
D
r
1
2

2

,
`

.
|
+ (12)

from which it is evident that
r
g should be non-vanishing, for the circuit to admit
5
a pair of roots with a positive real part tending to zero as .
We now consider the differential oscillator shown in Figure 3, which is
basically equivalent to the one in Figure 1, the only difference being that the
active part is made up of a nMOS pair.
As in the previous case, the nonlinear analysis of this circuit can be based on
the incremental circuit shown in Figure 1 (b), which is now obtained considering
the nonlinear characteristic ( ) v i
d
around the quiescent operating point,
2
0 0 2 1
I i i i
D D D
, of the composite one-port formed by the pair of nMOS
M
1
and M
2
and the current source
0
I . If the voltage v lies within the range
TN TN
V v V , the two active devices operate in pinch-off region and the
drain current
d
i can be obtained by imposing that the sum of the drain currents of
M
1
and M
2
are equal to
0
I , resulting

2
2
0
1
n
n
d
V
v
V
v
I i (13)
n n
k I V
0
2 and ( ) L W k k
n n
2 .
If, instead, the voltage v is greater than
TN
V , then M
2
goes into the triode region
and the drain current of M
1
, as it is easy to show, is described by the equation

( )
( )
]
]
]
]

,
`

.
|

2 2
2
2
2
0
2
2
0
2 1 1
v V
V v
V
v
V
v
I V v
V
I
i
n
TN
n n
TN
n
d
. (14)

On the other hand, if v is lower than
TN
V , then M
1
goes into the triode region
and an expression similar to (14) is obtained for the drain current of M
1
.
Therefore, the circuit equation is
( ) v i D
C
v
C L
D
C R
D
d
1 1 1
2

,
`

.
|
+ + (15)
where
6
( )
( )
( )

'

]
]
]
]

,
`

.
|

,
`

.
|



]
]
]
]

+
+

,
`

.
|

,
`

.
|
+

TN
n
TN
n
n n
TN
TN TN
n
n
TN
n
TN
n
n n
TN
d
V v
v V
V v
V
v
V
v
I
V
V v
I
V v V
V
v
V
v
I
V v
v V
V v
V
v
V
v
I
V
V v
I
v i
2 1 1
1
2 1 1
2 2
2
2
2
0
2
0
2
2
0
2 2
2
2
2
0
2
0
(16)

which has the derivative at the origin

0
'
2
2
) ( I k g
g
v i g
n m
m
d
. (17)

As in the previous circuit, equation (15) can be formulated as a regular
perturbation problem by introducing the parameter as in (5), and using the
dimensionless variables t t
0
,
n
V v x , and
n TN TN
V V x . Thus,
equation (15) can be put in the form (6) with 2
0 n
k I R k and
( )
( )
( )
( )
( )

'

+
+ +

.
1
2 1 1
1

1
2 1 1
2
2
2 2
2
2
2
2 2
TN
TN
TN
TN TN
TN
TN
TN
x x
x
x x
x x x x
x x x x x
x x
x
x x
x x x x
x n (18)

Observing that the characteristic ( ) v i
d
, whose diagram is shown in Figure 5,
exhibits at the origin the negative differential conductance g and that
2
m
g R k , equation (15) can be written as (11) and, consequently, in the form
(12) with ( ) x x n x f + ) ( .
Note that, the two circuits are described through formally identical equations,
but different in the nonlinearty ) (x f . The latter, which does not have a linear
7
part, is usually neglected although it makes the analysis of these circuits radically
different from the linear one, as it determines the existence of an oscillation, its
stability and shape. This depends on the parameter , which enables the
oscillator circuit to be studied as a perturbation of a purely harmonic oscillator
obtained for tending to zero. The value of measures how much the oscillator
behavior differs from this ideal behavior, and may not be very small when a fully
integrated solution is used due to the low quality factor of the inductor.


III. PERTURBATION SOLUTION METHOD
The oscillators considered above belong to the class of the so-called negative
differential-resistance oscillators [9],[10], whose incremental circuit around the
equilibrium state is shown in a generalized form in Figure 5 (a). We assume that
the linear part of nonlinearity is attributed to the linear circuit and we denote by
y and x the resulting one-port variables. The nonlinear part of circuit is
described by the function ( ) x f y , ( ) ( ) ( ) 0 0 0
'
f f , the one-port variables
being related through the transfer function ( ) ( ) ( ) s P s M y x s T of the
overall linear part of circuit. The nonlinear differential equation governing the
behavior of such system (Figure 5 (b)), i.e. ( ) ( ) ( ) x f D M x D P , can be written
in the normalized perturbation form ( ) ( ) ( ) x f D M x D N , , [8]. In the
particular case of second-order oscillators, to which here we specifically refer to,
this equation reduces to

( ) ( ) ( ) x f k D k x D D
0 1
2
+ + (19)

where and are analytical function of , that is ( ) L + + +
2
2 1
1 ,
( ) L + + +
2
2 1 0
.
Equation (19) represents the most general equation of second-order oscillators,
and has a form suitable for applying perturbation methods [11], as it enables the
oscillator circuit to be studied as a perturbation of a linearized feedback loop.
8
Accordingly, the periodic oscillation ( ) , t x of (19) is studied as a perturbation of
the generating solution, t A x cos
0 0
, of the linear limit problem ( ) 0 1
2
+ x D ,
obtained for 0 . This solution is sought in the form

( ) Re ,
]
]
]
]

+

m
jm
m
j
e H Ae t x

(20)

where A,
m
H , and are defined as

L
L L
+ + +
+ + + + +
2
2
1 0
2
2 , 1 ,
2
2 1 0


dt
d
H H H A A A A
m m m
(21)
and can be determined through the asymptotic method discussed in [8]. Here, we
develop a method for calculating both the steady-state and the transient
oscillation, using the basic idea underlying the method discussed in [12], whose
application to second-order oscillators in not straightforward.
As during the transient the amplitude, the harmonics and the frequency of
oscillation change in a relatively slow fashion with respect to the fast variations
of x in t , we assume that the coefficients of the power series in for A,
m
H
and are functions of the slow variable t .
To solve (19), the nonlinearity ( ) x f is expanded in a power series in

( ) ( ) L L + + + + +
1 0 2
2
1 0
f f x x x f x f (22)

where
k
x is the sum of the coefficients of the terms in
k
of (20). As the
coefficients
n
f are 2 -periodic in , they are developed as a Fourier series

( )
]
]
]
]

+ +

m
jm
m
j
e F e F F x f

1 0
Re (23)
L + + +
2
2 , 1 , 0 ,

m m m m
F F F F
9

2
0
, 0

2
1
d f F
n n
,

d e f F
jm
n n m

2
0
,

1
.

Moreover, the derivatives with respect to t that the linear operators ) , ( D N and
) , ( D M involve in equation (19), are carried out according to the equation
( ) ( )
( )
( )


jm
n m
n m jm jm
n m
e H m j
d
dH
e e H
dt
d
,
,
,
+ (24)

which allows us to exploit the method of complex amplitudes. By imposing that
the coefficients of terms with the same power of and with the same harmonic
order must balance in (19), we get the equations of the successive
approximations of (19). Balancing the zero-order terms in , we get ( ) 1
0
.
Balancing the first-order terms in , we get the determining equations of the first
approximation
( )
0 0 0 , 1 1
0
2
1
A F k
d
dA

+ (25)

0 , 1
0
0 1
1
2 2
F
A
k


(26)

0 ,
2
1 0
1 ,
1
m m
F
m
k m j k
H

+
(27)
where the
n m
F
,
s are defined as in (23).
The procedure may be continued in the same way to attain higher-order
approximations, that is to order ) (
3
O or higher [12], needed when the
parameter is not small enough or higher accuracy is desired. Here, we limit
ourselves to observe that when 0
1
and 0
0
k , as it often happens in
practical circuits, the second-order approximation to is needed to evaluate the
frequency deviation from
0
which in this case, as is readily shown, has the
following explicit form
]
]
]
]

+ +
2
0
2
0
0
0 , 1
1 0 2 1 , 1 1
0
2
Im
2
1


d
A d
d
dA
d
dF
k A R k
A
(28)
10
( )
( ) ( )
( )

+

2
1 ,
1
1
1
1 1 , 1

s
s s s
H C C R ,
( )
( )
( )
( )

2
0
cos
1
cos
2
1
0
d k
dx
x df
C
A x
k


involving calculations in the first iteration only. The calculation of the second-
order approximation are then completed determining
1
A and
2 , m
H .
It is worth noting that (28) shows how harmonics affect the shift in frequency
from
0
of an oscillator, which is minimized if the harmonics
2 , m
H are small.
This effect, first pointed out by Groszkowski [9],[13], is clearly highlighted by
(28) for the transient, too.
Not only does the first-order equation (25) make it possible to find the limit
cycle, but it also allows us to analyze the stability of the limit-cycle itself
(
*
0 0
A A ) and of the equilibrium state ( 0
0
A ), which are determined as
solutions of the equation
0
0 0 0 , 1 1
+ A F k . (29)
It can be found that the inequalities

*
0 0
0
0 , 1
1 0
A A
dA
dF
k

< (30)

0
0
0 , 1
1 0
0

>
A
dA
dF
k (31)
ensure the asymptotic stability of the limit-cycle and the instability of the
equilibrium state, respectively.
Note that inequality (31), referred to as the start-up condition [10], allows us only
to design the circuit so that the equilibrium point is unstable. Our analysis,
instead, not only allows us to ensure instability of the equilibrium point, but it
also enables us to determine the stability of the limit cycle and to effectively
construct it. It is worth noting that the existence of the periodic solution of (29) is
a necessary, but not sufficient condition, for the periodic oscillation to be
realized. For this, condition (30) must be fulfilled.

IV. APPLICATION EXAMPLE
11
We now show the application of the described method to the oscillator circuit
shown in Figure 3, which, as shown in Section II, can be described by equation
(19) with 1 2
m
g R , 1 , 2
1 m
g R k , 0
0
k , ( ) ( ) x x n x f + , being
( ) x n defined as in (18) and
m
g as in (17).
The start-up condition (31) for the differential MOS VCO simply leads to
2
0
>
n
k I R (31)
which clearly shows the relationship between the active part of the circuit (
0
I
and
n
k ) and the circuit losses R.
The steady-state oscillation amplitude can be determined by solving the
following equation

0 , 1 0
2
F
g R
g R
A
m
m

(32)
while condition (30), which leads to

2
0
0 , 1
0
<
dA
dF
k I R
n
, (33)
allows the stability condition for the limit cycle (32) to be verified. Equations
(32)-(33) are useful to predict the amplitude of a realizable oscillation, as shown
below, and can be used to find the steady-state solution needed in phase noise
evaluation [3]. For the considered MOS VCO, the steady-state frequency
deviation becomes

0
1 , 1
2
4
Im

A
R
g R
m
(34)
which allows us to analyze the dependence of the oscillation frequency on the tail
current
0
I , through the coefficient
m
g and the harmonic amplitudes (
1 , 1
R ),
which still depend on
1
k and, hence, on
m
g . This dependence allows us to
rigorously evaluate the indirect frequency modulation (FM) due to the flicker
noise in the tail current
0
I [14].
Moreover, equation (25) allows us to calculate the transient behavior of the
oscillation amplitude as shown in Figure 6. The circuit parameters fixed for the
simulation ensure that the stationary equilibrium state is unstable and the limit
12
cycle is stable, as conditions (31) and (33) hold. These conditions and the
uniqueness of the limit cycle can be easily verified by plotting
( )
0 0 , 1 1 0
A F k A g as a function of
0
A (Figure 7 curve b)). We note that the
derivative in the origin is positive, which ensures the start-up condition. Further,
as there exists only one zero and the derivative in the zero crossing is negative,
the uniqueness of the limit cycle as well as the stability of the limit cycle are
ensured. Note also that the variation of
1
k modifies the zero crossing and,
eventually, the zero derivative, thus modifying the amplitude of the limit cycle
and, eventually, the start-up condition, as shown by curves a) and c) in Figure 7
for greater and lower values of
1
k .
With reference to the bipolar VCO, shown in Figure 1, similar considerations
hold for the start up condition, the steady-state amplitude evaluation, the transient
calculation, the stability verification and the frequency deviation estimation.
It is interesting to note that, in both cases of bipolar and CMOS VCOs, if the
odd nonlinear characteristic ( ) x f is approximated by a cubic polynomial

( )
3
x k x f
F
(35)
being
F
k the value that best fits the nonlinear transfer characteristic, we get a
very simple expression for

3
0 0 , 1
4
3
A k F
F
(36)

and by integrating equation (25), a closed form expression for both the transient
and the steady-state oscillation can be derived in first approximation, that is
( ) ( ) t t A t x cos
0 0
with

( )
( )t k
F
e C k k
k
t A
1
1
1
0
1
3
1
2

+

(37)

being ( ) ( ) 0 1 4 3
2
0 1 1
+ t x k k k C
F
. Equation (37) allows us to analyze the
13
influence of both active and passive circuit parameters on the VCOs dynamic
behavior, providing a useful tool in design problems.

V. CONCLUSIONS
We performed a nonlinear analysis of the differential VCOs by preliminarly
developing a suitable nonlinear perturbation model of them, obtained using the
standard model of the bipolar or MOS transistors. For the sake of generality, we
analyzed the wide class of negative resistance second-order oscillators, starting
from their general perturbation equation. A method for predicting the steady-state
and the transient oscillation was presented focusing, in particular, on the
determining equations for the first approximations to the amplitude, the
harmonics and the frequency deviation, due to the nonlinearity effect. The
asymptotic orbital stability is also established, thus thoroughly analyzing the
dynamics of RF VCOs. All oscillation characteristics and circuit parameters have
been linked by useful analytical relationships, which are desirable in design
problems.

REFERENCES
[1] C. Samori, A. Lacaita, F. Villa and F. Zappa, Spectrum folding of phase noise
in LC tuned oscillators, IEEE Trans. Circuits Syst. II , vol.45, pp.781-790, July
1998
[2] D. Ham and A. Hajimiri, Concepts and methods in optimization of integrated
LC VCOs, IEEE J. Solid-State Circuits, vol. 36, pp. 896 909, June 2001.
[3] A. Demir, A. Mehrotra, and J. Roychowdhury, Phase noise in oscillators: a
unifying theory and numerical methods for characterization, IEEE Trans.
Circuits Syst. I, vol.47, no.5, pp. 655 -674, May 2000.
[4] Q. Huang, Phase Noise to Carrier Ratio in LC Oscillators, IEEE Trans.
Circuits Syst. I, vol. 47, pp. 965-972, July 2000.
[5] A. Shibutani, T. Saba, S. Moro, and S. Mori, Transient response of Colpitts-
VCO and its effect on performance of PLL system, IEEE Trans. Circuits Syst. I,
vol. 45, pp. 717-725, July 1998.
14
[6] G. Sarafian and B. Z. Kaplan, A new approach to the modeling of the dynamics
of RF VCOs and some of its practical implications, IEEE Trans. Circuits Syst.
I, vol. 40, pp. 895-901, Dec. 1993.
[7] D. O. Pederson and K. Mayaram, Analog Integrated Circuits for Communication.
Boston, MA: Kluwer Academic Publ., 1991.
[8] A. Buonomo and C. Di Bello, Asymptotic formulas in nearly sinusoidal
nonlinear oscillators, IEEE Trans. Circuits Syst. I, vol. 43, pp. 953-963, Dec.
1996.
[9] K.K. Clarke and D.T. Hess, Communication Circuits: Analysis and Design.
Reading, MA: Addison Wesley, 1971.
[10] N. M. Nguyen and R. G. Meyer, Start-up and frequency stability in high-
frequency oscillators, IEEE J. Solid-State Circuits, vol. 37, pp. 810-820, May
1992.
[11] R. Grimshaw, Nonlinear Ordinary Differential Equations. New York: Blackwell
Scientific, 1990.
[12] A. Buonomo and A. Lo Schiavo, A method for analysing the transient and the
steady-state oscillations in third-order oscillators with shifting bias, Int. J. Circ.
Theory and Appl., vol. 29, pp. 469-486, July 2001.
[13] J. Groszkowski, Frequency of Self-Oscillation. New York, NY: Macmillan,
1964.
[14] J.J Rael and A.A. Abidi, Physical Processes of phase noise in LC oscillators,
IEEE Custom Integrated Circuit Conf., 2000, pp.569-572.
15

FIGURE CAPTIONS
Fig. 1 Differential LC oscillator biased by tail current (a) and its equivalent circuit
(b).
Fig. 2 ( ) v i characteristic for the composite one-port formed by the pair of BJT T
1

and T
2
in Fig.1( 99 . 0
F
, 5 . 0
R
, A e I
s
13 1 , mA I 4
0
, mV V
T
25 ).
Fig. 3 CMOS differential oscillator.
Fig. 4 ( ) v i characteristic for the composite one-port formed by the pair of MOS M
1

and M
2
in Fig. 3 (
2
/ 20 V A k , 480 L W , mA I 4
0
, V V
N T
5 . 0 ).
Fig. 5 Negative-resistance oscillator: a) one-port model, b) feedback model.
Fig. 6 Oscillation envelope for the circuit in Figure 3 calculated with: nH L 4 ,
pF C 9545 . 1 , 341 R , mA I 4
0
, V V
N T
5 . 0 , 480 L W ,
2
/ 20 V A k , V V
DD
3 .
Fig. 7 Nonlinear function ( )
0 0 , 1 1 0
A F k A g for the circuit in Figure 3, calculated
with: a) 15 . 1
1
k , b) 05 . 1
1
k and c) 95 . 0
1
k .


16



I
0
V
CC
v
v
P
L/2 L/2
2 C 2 C
i

T
1
T
2



i

R L

C
v


Figure 1



b)
a)
17

-0.1 -0.05 0 0.05 0.1 0.15
0
0.5
1
1.5
2
2.5
3
3.5
4
v [V]
i [mA]


Figure 2




I
0
V
dd
v
v
P
L/2 L/2
2 C 2 C
i=i
d
M
1 M
2
M
5
M
6
V
DD


Figure 3

18

-1.5 -1 -0.5 0 0.5 1 1.5
0
0.5
1
1.5
2
2.5
3
3.5
4
i [mA]
v [V]


Figure 4



i

v


Frequency
dependent
linear
circuit


Active
resistive
nonlinear
circuit






T(s)
x

y


f(x)



Figure 5

b)
a)
19


0

50

100
150

200

-0.4
-0.2

0
0.2

0.4
[V]
time [ns]

Figure 6



0.1
0.2

0.3

0.4 0.5

0.6

-0.04

-0.03

-0.02

-0.01
0.01
0.02

0.03

0.04

A
0

k
1
F
10
- A
0

a)
b)
c)

Figure 7

You might also like