Information Technology Reference
In-Depth Information
3 Solving Circuits with Reduced SV Equations
The proposed method can be applied to circuits with non-linear elements such
as MOSFETs. After formulating the circuit matrix by the MNA method, apply
algorithm 1 to produce a reduced matrix. Then, it must be rearranged to the set
of explicit ordinary differential equations that is ready for the RK4 integration
method to evaluate. However, the derivation of the
A
sv
matrix is lengthy, for
instance, the entry
A
sv
(1
,
1) with the sum of conductance for the fulladder circuit
shown in Fig.1 is (
G
dsp
1
+
G
dsn
4
+
G
dsp
2
+
G
dsn
5
−
(
G
dsp
2
+
G
gsp
2
)
...
For
the sake of simplicity, let
A
sv
(1
,
1) be written as
G
1
, the sum of conductances,
G
2
=
A
sv
(1
,
2) and so on. For
B
sv
,let
i
1
be the sum of currents relating to the
first state variable, etc. Hence,
A
sv
X
1
=
B
sv
is shown as:
G
1
+
C
1
dt
(
G
dsp
2
∗
v
n
+1
1
v
n
+1
2
=
C
1
dt
v
1
+
i
1
C
2
G
2
(12)
G
4
+
C
2
dt
dt
v
2
+
i
2
G
3
The sizes of
A
sv
and
B
sv
depend on the number of dynamic elements attached
to the network. There must be at least one dynamic element in the circuit when
using the approach. For
N
elements in the circuit,
A
sv
and
B
sv
are
N
×
N
and
1. The matrix is then transformed to state variable form
dv
1
dt
=
f
(
v
1
,t
)
,
dv
2
dt
N
×
=
f
(
v
2
,t
). The first equation is written as
G
1
+
C
1
dt
v
n
+1
=
C
1
+
G
2
v
n
+1
v
1
+
i
1
dt
×
(13)
1
2
Rearranging gives
+
C
1
dt
v
1
+
dv
1
+
G
2
v
2
n
+1
=
C
1
G
1
v
n
+1
dt
v
1
+
i
1
(14)
1
The final form of the ODE for
v
1
is
dv
1
dt
G
1
v
n
+1
G
2
v
n
+1
=
i
1
−
−
1
2
(15)
C
1
The second equation can be written in the same way to become
dv
2
dt
G
3
v
n
+1
G
4
v
n
+1
=
i
2
−
−
1
2
(16)
C
2
From equations (15) and (16), it can be concluded that for the circuit equations
with
M
capacitors and hence
M
state variables, the matrix can be written as
dv
1
dt
dt
=
f
(
v
2
,t
)
...
dv
dt
=
f
(
v
M
,t
). The final form of the set of
ODE equations representing the non-linear circuit appears as:
=
f
(
v
1
,t
)
,
dv
2
=
i
1
−G
1
(
v
n
+1
)
−G
2
(
v
n
+1
2
)
···−G
M
(
v
n
+
M
)
dv
1
dt
1
C
1
=
i
2
−G
M
+1
(
v
n
+1
)
−G
M
+2
(
v
n
+1
2
)
···−G
M
+
M
(
v
n
+
M
)
dv
2
dt
1
C
2
(17)
.
=
i
M
−G
(
M−
1)
×M
+1
(
v
n
+1
)
···−G
(
M−
1)
×M
+
M
(
v
n
+
M
)
C
M
dv
M
dt
1
where
n
again is the number of time steps. These equations can be solved by the
Runge-Kutta method with Newton Raphson iteration.